Search Results

Search found 8687 results on 348 pages for 'per ersson'.

Page 39/348 | < Previous Page | 35 36 37 38 39 40 41 42 43 44 45 46  | Next Page >

  • Compare Your Internet Cost and Speed to Global Averages [Infographic]

    - by ETC
    Internet pricing and speed varies wildly across the world. The US, for instance, currently ranks 15th in speed but enjoys reasonably priced internet access. How reasonably priced? If you’re a US citizen you likely have an average internet access speed of 4.8 mbps and you pay a little over $3 per mbps. If you’re in Sweden, however, you likely have an 18 mbps connection and you pay a scant 63 cents per mpbs. The real envy of the internet speed Olympics by far is Japan with a mighty 61 mbps at a mere 27 cents per mbps. Hit up the link below for the full infographic (or use this local mirror if you need to dodge a firewall), then sound off in the comments with how you compare on the international scale. Internet Speeds and Costs Around the World [via Daily Infographic] Latest Features How-To Geek ETC Should You Delete Windows 7 Service Pack Backup Files to Save Space? What Can Super Mario Teach Us About Graphics Technology? Windows 7 Service Pack 1 is Released: But Should You Install It? How To Make Hundreds of Complex Photo Edits in Seconds With Photoshop Actions How to Enable User-Specific Wireless Networks in Windows 7 How to Use Google Chrome as Your Default PDF Reader (the Easy Way) Manage Your Favorite Social Accounts in Chrome and Iron with Seesmic E.T. II – Extinction [Fake Movie Sequel Video] Remastered King’s Quest Games Offer Classic Gaming on Modern Machines Compare Your Internet Cost and Speed to Global Averages [Infographic] Orbital Battle for Terra Wallpaper WizMouse Enables Mouse Over Scrolling on Any Window

    Read the article

  • How to improve batching performance

    - by user4241
    Hello, I am developing a sprite based 2D game for mobile platform(s) and I'm using OpenGL (well, actually Irrlicht) to render graphics. First I implemented sprite rendering in a simple way: every game object is rendered as a quad with its own GPU draw call, meaning that if I had 200 game objects, I made 200 draw calls per frame. Of course this was a bad choice and my game was completely CPU bound because there is a little CPU overhead assosiacted in every GPU draw call. GPU stayed idle most of the time. Now, I thought I could improve performance by collecting objects into large batches and rendering these batches with only a few draw calls. I implemented batching (so that every game object sharing the same texture is rendered in same batch) and thought that my problems are gone... only to find out that my frame rate was even lower than before. Why? Well, I have 200 (or more) game objects, and they are updated 60 times per second. Every frame I have to recalculate new position (translation and rotation) for vertices in CPU (GPU on mobile platforms does not support instancing so I can't do it there), and doing this calculation 48000 per second (200*60*4 since every sprite has 4 vertices) simply seems to be too slow. What I could do to improve performance? All game objects are moving/rotating (almost) every frame so I really have to recalculate vertex positions. Only optimization that I could think of is a look-up table for rotations so that I wouldn't have to calculate them. Would point sprites help? Any nasty hacks? Anything else? Thanks.

    Read the article

  • How do I align my partition table properly?

    - by Jorge Castro
    I am in the process of building my first RAID5 array. I've used mdadm to create the following set up: root@bondigas:~# mdadm --detail /dev/md1 /dev/md1: Version : 00.90 Creation Time : Wed Oct 20 20:00:41 2010 Raid Level : raid5 Array Size : 5860543488 (5589.05 GiB 6001.20 GB) Used Dev Size : 1953514496 (1863.02 GiB 2000.40 GB) Raid Devices : 4 Total Devices : 4 Preferred Minor : 1 Persistence : Superblock is persistent Update Time : Wed Oct 20 20:13:48 2010 State : clean, degraded, recovering Active Devices : 3 Working Devices : 4 Failed Devices : 0 Spare Devices : 1 Layout : left-symmetric Chunk Size : 64K Rebuild Status : 1% complete UUID : f6dc829e:aa29b476:edd1ef19:85032322 (local to host bondigas) Events : 0.12 Number Major Minor RaidDevice State 0 8 16 0 active sync /dev/sdb 1 8 32 1 active sync /dev/sdc 2 8 48 2 active sync /dev/sdd 4 8 64 3 spare rebuilding /dev/sde While that's going I decided to format the beast with the following command: root@bondigas:~# mkfs.ext4 /dev/md1p1 mke2fs 1.41.11 (14-Mar-2010) /dev/md1p1 alignment is offset by 63488 bytes. This may result in very poor performance, (re)-partitioning suggested. Filesystem label= OS type: Linux Block size=4096 (log=2) Fragment size=4096 (log=2) Stride=16 blocks, Stripe width=48 blocks 97853440 inodes, 391394047 blocks 19569702 blocks (5.00%) reserved for the super user First data block=0 Maximum filesystem blocks=0 11945 block groups 32768 blocks per group, 32768 fragments per group 8192 inodes per group Superblock backups stored on blocks: 32768, 98304, 163840, 229376, 294912, 819200, 884736, 1605632, 2654208, 4096000, 7962624, 11239424, 20480000, 23887872, 71663616, 78675968, 102400000, 214990848 Writing inode tables: ^C 27/11945 root@bondigas:~# ^C I am unsure what to do about "/dev/md1p1 alignment is offset by 63488 bytes." and how to properly partition the disks to match so I can format it properly.

    Read the article

  • Magento - How to manage multiple base currencies and multiple payment gateways?

    - by Diego
    I have two requirements to satisfy, I hope someone with more experience can help me sorting them out. Multiple Base Currencies My client wants to allow visitors to place orders in whatever currency they prefer, choosing from the ones he’ll configure. Magento only supports one Base Currency, and this is, obviously, not what I need. I checked the solution involving multiple websites, but I need a customer to be registered once and stay on the same website, not to switch from one to the other and have to register/log in on each. Manage multiple Payment Gateways per currency and per payment method This is another crucial requirement, and it’s tied to the first one. My client wants to “route” payments in different currencies to different accounts. He’ll thus have one for Euro, one for USD and one for GBP. Whenever a customer pays with one of these currencies, the payment gateway has to be chosen accordingly. Additionally, the gateway should be different depending on other rules. For example, if customer pays with a Debit Card, my client will have a payment gateway configured especially for it. If customer pays with MasterCard, the gateway will be different, and so on. The complication, in this case, arises from the fact that my client uses Realex Payments and, although it would be possible for him to open multiple accounts, the Realex module expects one single gateway. In a normal scenario, we would need up to six instead: Payment with Debit Card in Euro Payment with Credit Card in Euro Payment with Debit Card in US Dollars Payment with Credit Card in US Dollars Payment with Debit Card in GB Pounds Payment with Credit Card in GB Pounds This, of course, if he doesn’t decide to accept other payment methods, such as bank transfer, which would add one more gateway per currency. Is there a way to achieve the above in Magento? I never had such complicated requirements before, and I’m a bit lost. Thanks in advance for the help.

    Read the article

  • MightyMintyBoost Is a 3-in-1 Gadget Charger

    - by ETC
    If you’re looking for a versatile battery booster, this DIY 3-in-1 solar/usb/wall current charger known as the MightyMintyBoost will top of your phone, mp3 player, and other gadgets with ease. Instructables user Honus didn’t just build the MightMintyBoost to geek out and show off his electronics project skills (although it’s certainly a nifty little project to do so), he’s serious about solar power and the impact clean energy has: Apple has sold over 30 million iPodTouch/iPhone units- imagine charging all of them via solar power…. If every iPhone/iPodTouch sold was fully charged every day (averaging the battery capacity) via solar power instead of fossil fuel power we would save approximately 50.644gWh of energy, roughly equivalent to 75,965,625 lbs. of CO2 in the atmosphere per year. Granted that’s a best case scenario (assuming you can get enough sunlight per day and approximately 1.5 lbs. CO2 produced per kWh used.) Of course, that doesn’t even figure in all the other iPods, cell phones, PDAs, microcontrollers (I use it to power my Arduino projects) and other USB devices that can be powered by this charger- one little solar cell charger may not seem like it can make a difference but add all those millions of devices together and that’s a lot of energy! His MightyMintyBoost is a battery booster for devices that can charge via USB and it accepts incoming current from the solar panel on top (or, on cloudy days can be charged via a wall charger or the USB port on your computer). Hit up the link below to see his full build guide and create your own MightyMintyBoost. MightyMintyBoost [Instructables] Latest Features How-To Geek ETC Internet Explorer 9 RC Now Available: Here’s the Most Interesting New Stuff Here’s a Super Simple Trick to Defeating Fake Anti-Virus Malware How to Change the Default Application for Android Tasks Stop Believing TV’s Lies: The Real Truth About "Enhancing" Images The How-To Geek Valentine’s Day Gift Guide Inspire Geek Love with These Hilarious Geek Valentines MyPaint is an Open-Source Graphics App for Digital Painters Can the Birds and Pigs Really Be Friends in the End? [Angry Birds Video] Add the 2D Version of the New Unity Interface to Ubuntu 10.10 and 11.04 MightyMintyBoost Is a 3-in-1 Gadget Charger Watson Ties Against Human Jeopardy Opponents Peaceful Tropical Cavern Wallpaper

    Read the article

  • Big Data Appliance X4-2 Release Announcement

    - by Jean-Pierre Dijcks
    Today we are announcing the release of the 3rd generation Big Data Appliance. Read the Press Release here. Software Focus The focus for this 3rd generation of Big Data Appliance is: Comprehensive and Open - Big Data Appliance now includes all Cloudera Software, including Back-up and Disaster Recovery (BDR), Search, Impala, Navigator as well as the previously included components (like CDH, HBase and Cloudera Manager) and Oracle NoSQL Database (CE or EE). Lower TCO then DIY Hadoop Systems Simplified Operations while providing an open platform for the organization Comprehensive security including the new Audit Vault and Database Firewall software, Apache Sentry and Kerberos configured out-of-the-box Hardware Update A good place to start is to quickly review the hardware differences (no price changes!). On a per node basis the following is a comparison between old and new (X3-2) hardware: Big Data Appliance X3-2 Big Data Appliance X4-2 CPU 2 x 8-Core Intel® Xeon® E5-2660 (2.2 GHz) 2 x 8-Core Intel® Xeon® E5-2650 V2 (2.6 GHz) Memory 64GB 64GB Disk 12 x 3TB High Capacity SAS 12 x 4TB High Capacity SAS InfiniBand 40Gb/sec 40Gb/sec Ethernet 10Gb/sec 10Gb/sec For all the details on the environmentals and other useful information, review the data sheet for Big Data Appliance X4-2. The larger disks give BDA X4-2 33% more capacity over the previous generation while adding faster CPUs. Memory for BDA is expandable to 512 GB per node and can be done on a per-node basis, for example for NameNodes or for HBase region servers, or for NoSQL Database nodes. Software Details More details in terms of software and the current versions (note BDA follows a three monthly update cycle for Cloudera and other software): Big Data Appliance 2.2 Software Stack Big Data Appliance 2.3 Software Stack Linux Oracle Linux 5.8 with UEK 1 Oracle Linux 6.4 with UEK 2 JDK JDK 6 JDK 7 Cloudera CDH CDH 4.3 CDH 4.4 Cloudera Manager CM 4.6 CM 4.7 And like we said at the beginning it is important to understand that all other Cloudera components are now included in the price of Oracle Big Data Appliance. They are fully supported by Oracle and available for all BDA customers. For more information: Big Data Appliance Data Sheet Big Data Connectors Data Sheet Oracle NoSQL Database Data Sheet (CE | EE) Oracle Advanced Analytics Data Sheet

    Read the article

  • Se non ti sei unito alla Customer Experience Revolution? Il materiale è tutto qui!

    - by Silvia Valgoi
    Se ti sei perso questo interesante Executive workshop, non preoccuparti, qui puoi trovare gli interventi dei relatori.Durante l'evento Oracle, Accenture ed il professor Enrico Finzi hanno condiviso l'approccio alla Customer Experience vista come strategia per dare vita a processi più completi ed innovativi, per generare e gestire l’interazione con i consumatori, su tutti i canali. E' stato un momento importante per: comprendere perché la Customer Experience è diventata la componente più importante e strategica del tuo business scoprire come la Customer Experience accelleri l’acquisizione di nuovi clienti, incrementi la fidelizzazione ad un brand/prodotto/servizio, migliori l’efficienza operativa e sostenga le vendite conoscere come le soluzioni di Customer Experience possono aiutare le aziende a far vivere questa esperienza in modo coerente, personalizzata, attraverso tutti i canali e su tutti i dispositivi, ottenendo risultati misurabile Ecco le presentazioni e i video presentati durante i lavori: &amp;lt;p&amp;gt; &amp;lt;/p&amp;gt; Oracle Customer Experience - Empowering People. Powering Brands - Armando Janigro, Sales Development Manager, Oracle         How to win with Customer Experience - Nadia Dallafiore, Senior Manager CRM Retail  Accenture   Customer Experience e selezione Darwiniana della marca - Enrico Finzi, Sociologo, Presidente AstraRicerche   Engage.Win.Develop.Keep LinkedIn: Customer Concepts Exchange Facebook: Oracle Customer Experience

    Read the article

  • Parallel Classloading Revisited: Fully Concurrent Loading

    - by davidholmes
    Java 7 introduced support for parallel classloading. A description of that project and its goals can be found here: http://openjdk.java.net/groups/core-libs/ClassLoaderProposal.html The solution for parallel classloading was to add to each class loader a ConcurrentHashMap, referenced through a new field, parallelLockMap. This contains a mapping from class names to Objects to use as a classloading lock for that class name. This was then used in the following way: protected Class loadClass(String name, boolean resolve) throws ClassNotFoundException { synchronized (getClassLoadingLock(name)) { // First, check if the class has already been loaded Class c = findLoadedClass(name); if (c == null) { long t0 = System.nanoTime(); try { if (parent != null) { c = parent.loadClass(name, false); } else { c = findBootstrapClassOrNull(name); } } catch (ClassNotFoundException e) { // ClassNotFoundException thrown if class not found // from the non-null parent class loader } if (c == null) { // If still not found, then invoke findClass in order // to find the class. long t1 = System.nanoTime(); c = findClass(name); // this is the defining class loader; record the stats sun.misc.PerfCounter.getParentDelegationTime().addTime(t1 - t0); sun.misc.PerfCounter.getFindClassTime().addElapsedTimeFrom(t1); sun.misc.PerfCounter.getFindClasses().increment(); } } if (resolve) { resolveClass(c); } return c; } } Where getClassLoadingLock simply does: protected Object getClassLoadingLock(String className) { Object lock = this; if (parallelLockMap != null) { Object newLock = new Object(); lock = parallelLockMap.putIfAbsent(className, newLock); if (lock == null) { lock = newLock; } } return lock; } This approach is very inefficient in terms of the space used per map and the number of maps. First, there is a map per-classloader. As per the code above under normal delegation the current classloader creates and acquires a lock for the given class, checks if it is already loaded, then asks its parent to load it; the parent in turn creates another lock in its own map, checks if the class is already loaded and then delegates to its parent and so on till the boot loader is invoked for which there is no map and no lock. So even in the simplest of applications, you will have two maps (in the system and extensions loaders) for every class that has to be loaded transitively from the application's main class. If you knew before hand which loader would actually load the class the locking would only need to be performed in that loader. As it stands the locking is completely unnecessary for all classes loaded by the boot loader. Secondly, once loading has completed and findClass will return the class, the lock and the map entry is completely unnecessary. But as it stands, the lock objects and their associated entries are never removed from the map. It is worth understanding exactly what the locking is intended to achieve, as this will help us understand potential remedies to the above inefficiencies. Given this is the support for parallel classloading, the class loader itself is unlikely to need to guard against concurrent load attempts - and if that were not the case it is likely that the classloader would need a different means to protect itself rather than a lock per class. Ultimately when a class file is located and the class has to be loaded, defineClass is called which calls into the VM - the VM does not require any locking at the Java level and uses its own mutexes for guarding its internal data structures (such as the system dictionary). The classloader locking is primarily needed to address the following situation: if two threads attempt to load the same class, one will initiate the request through the appropriate loader and eventually cause defineClass to be invoked. Meanwhile the second attempt will block trying to acquire the lock. Once the class is loaded the first thread will release the lock, allowing the second to acquire it. The second thread then sees that the class has now been loaded and will return that class. Neither thread can tell which did the loading and they both continue successfully. Consider if no lock was acquired in the classloader. Both threads will eventually locate the file for the class, read in the bytecodes and call defineClass to actually load the class. In this case the first to call defineClass will succeed, while the second will encounter an exception due to an attempted redefinition of an existing class. It is solely for this error condition that the lock has to be used. (Note that parallel capable classloaders should not need to be doing old deadlock-avoidance tricks like doing a wait() on the lock object\!). There are a number of obvious things we can try to solve this problem and they basically take three forms: Remove the need for locking. This might be achieved by having a new version of defineClass which acts like defineClassIfNotPresent - simply returning an existing Class rather than triggering an exception. Increase the coarseness of locking to reduce the number of lock objects and/or maps. For example, using a single shared lockMap instead of a per-loader lockMap. Reduce the lifetime of lock objects so that entries are removed from the map when no longer needed (eg remove after loading, use weak references to the lock objects and cleanup the map periodically). There are pros and cons to each of these approaches. Unfortunately a significant "con" is that the API introduced in Java 7 to support parallel classloading has essentially mandated that these locks do in fact exist, and they are accessible to the application code (indirectly through the classloader if it exposes them - which a custom loader might do - and regardless they are accessible to custom classloaders). So while we can reason that we could do parallel classloading with no locking, we can not implement this without breaking the specification for parallel classloading that was put in place for Java 7. Similarly we might reason that we can remove a mapping (and the lock object) because the class is already loaded, but this would again violate the specification because it can be reasoned that the following assertion should hold true: Object lock1 = loader.getClassLoadingLock(name); loader.loadClass(name); Object lock2 = loader.getClassLoadingLock(name); assert lock1 == lock2; Without modifying the specification, or at least doing some creative wordsmithing on it, options 1 and 3 are precluded. Even then there are caveats, for example if findLoadedClass is not atomic with respect to defineClass, then you can have concurrent calls to findLoadedClass from different threads and that could be expensive (this is also an argument against moving findLoadedClass outside the locked region - it may speed up the common case where the class is already loaded, but the cost of re-executing after acquiring the lock could be prohibitive. Even option 2 might need some wordsmithing on the specification because the specification for getClassLoadingLock states "returns a dedicated object associated with the specified class name". The question is, what does "dedicated" mean here? Does it mean unique in the sense that the returned object is only associated with the given class in the current loader? Or can the object actually guard loading of multiple classes, possibly across different class loaders? So it seems that changing the specification will be inevitable if we wish to do something here. In which case lets go for something that more cleanly defines what we want to be doing: fully concurrent class-loading. Note: defineClassIfNotPresent is already implemented in the VM as find_or_define_class. It is only used if the AllowParallelDefineClass flag is set. This gives us an easy hook into existing VM mechanics. Proposal: Fully Concurrent ClassLoaders The proposal is that we expand on the notion of a parallel capable class loader and define a "fully concurrent parallel capable class loader" or fully concurrent loader, for short. A fully concurrent loader uses no synchronization in loadClass and the VM uses the "parallel define class" mechanism. For a fully concurrent loader getClassLoadingLock() can return null (or perhaps not - it doesn't matter as we won't use the result anyway). At present we have not made any changes to this method. All the parallel capable JDK classloaders become fully concurrent loaders. This doesn't require any code re-design as none of the mechanisms implemented rely on the per-name locking provided by the parallelLockMap. This seems to give us a path to remove all locking at the Java level during classloading, while retaining full compatibility with Java 7 parallel capable loaders. Fully concurrent loaders will still encounter the performance penalty associated with concurrent attempts to find and prepare a class's bytecode for definition by the VM. What this penalty is depends on the number of concurrent load attempts possible (a function of the number of threads and the application logic, and dependent on the number of processors), and the costs associated with finding and preparing the bytecodes. This obviously has to be measured across a range of applications. Preliminary webrevs: http://cr.openjdk.java.net/~dholmes/concurrent-loaders/webrev.hotspot/ http://cr.openjdk.java.net/~dholmes/concurrent-loaders/webrev.jdk/ Please direct all comments to the mailing list [email protected].

    Read the article

  • Data Pump: Consistent Export?

    - by Mike Dietrich
    Ouch ... I have to admit as I did say in several workshops in the past weeks that a data pump export with expdp is per se consistent. Well ... I thought it is ... but it's not. Thanks to a customer who is doing a large unicode migration at the moment. We were discussing parameters in the expdp's par file. And I did ask my colleagues after doing some research on MOS. And here are the results of my "research": MOS Note 377218.1 has a nice example showing a data pump export of a partitioned table with DELETEs on that table as inconsistent Background:Back in the old 9i days when Data Pump was designed flashback technology wasn't as popular and well known as today - and UNDO usage was the major concern as a consistent per default export would have heavily relied on UNDO. That's why - similar to good ol' exp - the export won't operate per default in consistency mode To get a consistent data pump export with expdp you'll have to set: FLASHBACK_TIME=SYSTIMESTAMPin your parameter file. Then it will be consistent according to the timestamp when the process has been started. You could use FLASHBACK_SCN instead and determine the SCN beforehand if you'd like to be exact. So sorry if I had proclaimed a feature which unfortunately is not there by default - Mike

    Read the article

  • NHibernate Tools: Visual NHibernate

    - by Ricardo Peres
    You probably know that I’m a big fan of Slyce Software’s Visual NHibernate. To me, it is the best tool for generating your entities and mappings from an existing database (it also allows you to go the other way, but I honestly have never used it that way). What I like most about it: Great support: folks at Slyce always listen to your suggestions, give you feedback in a timely manner, and I was even lucky enough to have some of my suggestions implemented! The templating engine, which is very powerful, and more user-friendly than, for example, MyGeneration’s; one of the included templates is Sharp Architecture; Advanced model validations: it even warns you about having lazy properties declared in non-lazy entities; Integration with NHibernate Validator and generation of validation rules automatically based on the database, or on user-defined model settings; The designer: they opted for not displaying all entities in a single screen, which I think was a good decision; has support for all inheritance strategies (table per class hierarchy, table per class, table per concrete class); Generation of FluentNHibernate mappings as well as hbm.xml. I could name others, but… why don’t you see for yourself? There is a demo version available for downloading. By the way, I am in no way related to Slyce, I just happen to like their software!

    Read the article

  • Introducing Oracle Multitenant

    - by OracleMultitenant
    0 0 1 1142 6510 Oracle Corporation 54 15 7637 14.0 Normal 0 false false false EN-US JA X-NONE /* Style Definitions */ table.MsoNormalTable {mso-style-name:"Table Normal"; mso-tstyle-rowband-size:0; mso-tstyle-colband-size:0; mso-style-noshow:yes; mso-style-priority:99; mso-style-parent:""; mso-padding-alt:0in 5.4pt 0in 5.4pt; mso-para-margin:0in; mso-para-margin-bottom:.0001pt; mso-pagination:widow-orphan; font-size:10.0pt; font-family:"Times New Roman"; mso-fareast-language:JA;} The First Database Designed for the Cloud Today Oracle announced the general availability (GA) of Oracle Database 12c, the first database designed for the Cloud. Oracle Multitenant, new with Oracle Database 12c, is a key component of this – a new architecture for consolidating databases and simplifying operations in the Cloud. With this, the inaugural post in the Multitenant blog, my goal is to start the conversation about Oracle Multitenant. We are very proud of this new architecture, which we view as a major advance for Oracle. Customers, partners and analysts who have had previews are very excited about its capabilities and its flexibility. This high level review of Oracle Multitenant will touch on our design considerations and how we re-architected our database for the cloud. I’ll briefly describe our new multitenant architecture and explain it’s key benefits. Finally I’ll mention some of the major use cases we see for Oracle Multitenant. Industry Trends We always start by talking to our customers about the pressures and challenges they’re facing and what trends they’re seeing in the industry. Some things don’t change. They face the same pressures and the same requirements as ever: Pressure to do more with less; be faster, leaner, cheaper, and deliver services 24/7. Big companies have achieved scale. Now they want to realize economies of scale. As ever, DBAs are faced with the challenges of patching and upgrading large numbers of databases, and provisioning new ones.  Requirements are familiar: Performance, scalability, reliability and high availability are non-negotiable. They need ever more security in this threatening climate. There’s no time to stop and retool with new applications. What’s new are the trends. These are the techniques to use to respond to these pressures within the constraints of the requirements. With the advent of cloud computing and availability of massively powerful servers – even engineered systems such as Exadata – our customers want to consolidate many applications into fewer larger servers. There’s a move to standardized services – even self-service. Consolidation Consolidation is not new; companies have tried various different approaches to consolidation of databases in the cloud. One approach is to partition a powerful server between several virtual machines, one per application. A downside of this is that you have the resource and management overheads of OS and RDBMS per VM – that is, per application. Another is that you have replaced physical sprawl with virtual sprawl and virtual sprawl is still expensive to manage. In the dedicated database model, we have a single physical server supporting multiple databases, one per application. So there’s a shared OS overhead, but RDBMS process and memory overhead are replicated per application. Let's think about our traditional Oracle Database architecture. Every time we create a database, be it a production database, a development or a test database, what do we do? We create a set of files, we allocate a bunch of memory for managing the data, and we kick off a series of background processes. This is replicated for every one of the databases that we create. As more and more databases are fired up, these replicated overheads quickly consume the available server resources and this limits the number of applications we can run on any given server. In Oracle Database 11g and earlier the highest degree of consolidation could be achieved by what we call schema consolidation. In this model we have one big server with one big database. Individual applications are installed in separate schemas or table-owners. Database overheads are shared between all applications, which affords maximum consolidation. The shortcomings are that application changes are often required. There is no tenant isolation. One bad apple can spoil the whole batch. New Architecture & Benefits In Oracle Database 12c, we have a new multitenant architecture, featuring pluggable databases. This delivers all the resource utilization advantages of schema consolidation with none of the downsides. There are two parts to the term “pluggable database”: "pluggable", which is new, and "database", which is familiar.  Before we get to the exciting new stuff let’s discuss what hasn’t changed. A pluggable database is a fully functional Oracle database. It’s not watered down in any way. From the perspective of an application or an end user it hasn’t changed at all. This is very important because it means that no application changes are required to adopt this new architecture. There are many thousands of applications built on Oracle databases and they are all ready to run on Oracle Multitenant. So we have these self-contained pluggable databases (PDBs), and as their name suggests, they are plugged into a multitenant container database (CDB). The CDB behaves as a single database from the operations point of view. Very much as we had with the schema consolidation model, we only have a single set of Oracle background processes and a single, shared database memory requirement. This gives us very high consolidation density, which affords maximum reduction in capital expenses (CapEx). By performing management operations at the CDB level – “managing many as one” – we can achieve great reductions in operating expenses (OpEx) as well, but we retain granular control where appropriate. Furthermore, the “pluggability” capability gives us portability and this adds a tremendous amount of agility. We can simply unplug a PDB from one CDB and plug it into another CDB, for example to move it from one SLA tier to another. I'll explore all these new capabilities in much more detail in a future posting.  Use Cases We can identify a number of use cases for Oracle Multitenant. Here are a few of the major ones. 0 0 1 113 650 Oracle Corporation 5 1 762 14.0 Normal 0 false false false EN-US JA X-NONE /* Style Definitions */ table.MsoNormalTable {mso-style-name:"Table Normal"; mso-tstyle-rowband-size:0; mso-tstyle-colband-size:0; mso-style-noshow:yes; mso-style-priority:99; mso-style-parent:""; mso-padding-alt:0in 5.4pt 0in 5.4pt; mso-para-margin:0in; mso-para-margin-bottom:.0001pt; mso-pagination:widow-orphan; font-size:10.0pt; font-family:"Times New Roman"; mso-fareast-language:JA;} Development / Testing where individual engineers need rapid provisioning and recycling of private copies of a few "master test databases" Consolidation of disparate applications using fewer, more powerful servers Software as a Service deploying separate copies of identical applications to individual tenants Database as a Service typically self-service provisioning of databases on the private cloud Application Distribution from ISV / Installation by Customer Eliminating many typical installation steps (create schema, import seed data, import application code PL/SQL…) - just plug in a PDB! High volume data distribution literally via disk drives in envelopes distributed by truck! - distribution of things like GIS or MDM master databases …various others! Benefits Previous approaches to consolidation have involved a trade-off between reductions in Capital Expenses (CapEx) and Operating Expenses (OpEx), and they’ve usually come at the expense of agility. With Oracle Multitenant you can have your cake and eat it: Minimize CapEx More Applications per server Minimize OpEx Manage many as one Standardized procedures and services Rapid provisioning Maximize Agility Cloning for development and testing Portability through pluggability Scalability with RAC Ease of Adoption Applications run unchanged It’s a pure deployment choice. Neither the database backend nor the application needs to be changed. In future postings I’ll explore various aspects in more detail. However, if you feel compelled to devour everything you can about Oracle Multitenant this very minute, have no fear. Visit the Multitenant page on OTN and explore the various resources we have available there. Among these, Oracle Distinguished Product Manager Bryn Llewellyn has written an excellent, thorough, and exhaustively detailed White Paper about Oracle Multitenant, which is available here.  Follow me  I tweet @OraclePDB #OracleMultitenant

    Read the article

  • How do I backup my customer's data?

    - by marcamillion
    If you run a SaaS app, or work on one, I would love to hear from you. Where the safety and security of your customer's data is paramount, how do you secure it and back it up? I would love to know your main host (e.g. Heroku, Engine Yard, Rackspace, MediaTemple, etc.) and who you use for your backup. Be as detailed as possible - e.g. a quick overview of your service and the data you store (images for instance), what happens with the images when the user uploads them (e.g. they go to your Linode VPS, and posted to the site for them to see - then they are automatically sent to AWS or wherever, then once a week they are backed up to tape by the managed hosting provider, and you also back them up to your house/office). If you could also give some idea as to what the unit cost (per GB/per user/per month) of storage is - on average, I would really appreciate that. Getting ready to launch my app, and I would love to get some more perspective on the nitty gritty details involved. Thanks!

    Read the article

  • How to customize web-app (pages and UI) for different customers

    - by demoncodemonkey
    We have an ASP.NET web-application which has become difficult to maintain, and I'm looking for ideas on how to redesign it. It's an employee administration system which can be highly customized for each of our customers. Let me explain how it works now: On the default page we have a menu where a user can select a task, such as Create Employee or View Timesheet. I'll use Create Employee as an example. When a user selects Create Employee from the menu, an ASPX page is loaded which contains a dynamically loaded usercontrol for the selected menuitem, e.g. for Create Employee this would be AddEmployee.ascx If the user clicks Save on the control, it navigates to the default page. Some menuitems involve multiple steps, so if the user clicks Next on a multi-step flow then it will navigate to the next page in the flow, and so on until it reaches the final step, where clicking Save navigates to the default page. Some customers may require an extra step in the Create Employee flow (e.g. SecurityClearance.ascx) but others may not. Different customers may use the same ASCX usercontrol, so in the AddEmployee.OnInit we can customize the fields for that customer, i.e. making certain fields hidden or readonly or mandatory. The following things are customizable per customer: Menu items Steps in each flow (ascx control names) Hidden fields in each ascx Mandatory fields in each ascx Rules relating to each ascx, which allows certain logic to be used in the code for that customer The customizations are held in a huge XML file per customer, which could be 7500 lines long. Is there any framework or rules-engine that we could use to customize our application in this way? How do other applications manage customizations per customer?

    Read the article

  • AdSense (reports) and custom channels

    - by RobbertT
    Please help me to further understand custom channels. As Google says it is a way to map your ads, but I still have a few questions: Is it correct that a single custom channel per 1 ad is not very useful, since you can specify Ad blocks in the AdSense reports? I have multiple Ads in multiple custom channels. After this I created 1 custom channel and added all the ads to it. I made this channel targetable, so people can target through this channel on all ads at once. Is this a good way to do it? In other words, is it possible to have ads in multiple custom channels (without targeting, just for analyzing) and then create 1 custom channel with targeting that embraces all the (desired) ads? Why is it not possible for me to analyze custom channels (or ad blocks & formats) per site in the Adsense (reports). Or am I doing something wrong? If not, I have to create different custom channels per site to see how certain ads are doing on a site level?

    Read the article

  • Web hosting company basically forces me to use their domain name [closed]

    - by Jinx
    I've recently stumbled upon an unusual problem with one of hosting companies called giga-international.com. Anyway, I've ordered com.hr domain from Croatian domain name registration company, and my client insisted on using this host provider as couple of his friends already are hosted with them. I thought something was fishy when the first result on Google for Giga International was this little forum rant instead of their webpage. When I was checking their services they listed many features etc... space available, bandwidth etc. I just wanted to check how much ram do I get for my PHP scripts so I emailed them, and they told me that was company secret. Seriously? Anyway, since my client still insisted on hosting with them I've bought their Webspace package. During registration I had to choose free domain name because I couldn't advance registration without it. Nowhere was said, not even in general terms and conditions that I wouldn't be able to change that domain name. At least not for double the price of domain name per year. They said I can either move my domain name over to them (and pay them domain registration), or pay them 1 Euro per month for managing a DNS entry. On any previous hosting solution I was able to manage my domain names just by pointing my domain to their name servers, and this is something completely new and absurd for me. They also said that usual approach is not possible because of security and hardware limitations. I'd like to know what you guys think about this case, and should I report, and where should I report this case. In short. They forced me to register free domain name which doesn't suit my needs in order to register for their webspace package, and refuse to change domain name for my account until I either transfer domain to them or pay them DNS management which costs double the price of the domain name per year.

    Read the article

  • Is it better to have multiple domains for cities or one single TLD?

    - by Brett
    I make websites for small businesses, and for some reason business owners love to have several domains with the same website but the TLD containing the city name. For example: 1. smallbizname.com 2. clevelandsmallbizname.com 3. columbussmallbizname.com 4. cincinnatismallbizname.com ... and so on. I've seen questions about localization per country aspects, but this is a much smaller scale, so I don't think the same rules apply. The problem I have is the companies never want to write separate content per domain, just have the same website hosted several times at each domain. I feel this probably hurts SEO for two reasons: 1. Traffic gets scattered throughout domains, could be boosting just one domain. 2. Duplicate content penalty because the content is identical. My question boils down to this... Should I redirect all the city domains to the main business name domain, or does having these separate sites help to rank better per city? And if they are redirected, how does google rank the redirects? Thanks for any input on this issue!!

    Read the article

  • How to develop a Windows 8 app in 30 days!

    - by Scott Spradlin
    Begin your 30-day journey to create a Windows Store style app. Sign up to get started and receive: Insider tips and tricks on Windows 8 application development. Personal on-the-phone access to a Windows 8 architect*. An exclusive one-on-one Windows Store design consultation*. An opportunity to get expert help from a Microsoft Services Engineer at an App Excellence Lab. Sign up today and get started. Your new Windows 8 app could be mere days away. * Offer good only to legal residents in the 50 United States & D.C., age 18 or older to hobbyists, professionals or developers in the field of software tech who sign up for building a Windows 8 application on www.generationapp.com. Offer limited to 250 design consultations per month and 500 technical review consultations per month, on a first come first served basis. Limit of one session of each offer type per person. This offer is non-transferable and cannot be combined with any other offer. This offer ends when supplies are exhausted, and is not redeemable for cash.

    Read the article

  • Server Setups for Agencies [closed]

    - by styks1987
    We are considering consolidating our server administration by cutting down on the number of systems we currently use to manage all our websites(~65 websites). Currently we have a testing server and 3 production servers. (2 - cherokee(linode), 1 - apache (mt)) We don't have a dedicated server admin, so I am stuck with managing all these servers, and as a developer, I don't want to deal with all the server hassle. So my main goal is to cut down on the time spent messing with the servers. We have looked at Pagoda Box and AppFog as possibilities. I am not sure if Pagoda Box would be cost effective. With 65+ websites we may end up paying anywhere from 0 to $50+ per website per month. Right now we page about $250 per month for the 4 VPS servers mentioned above. We already use Capistrano for deployment. I have the opportunity to completely overhaul the entire setup and I would like some feedback on where you found your information for large scale server management or how you currently do it. Articles are welcome. In summary: What is new (past 2 years) in simple server management arena? If you work at an agency or have had agency experience, how do/did you manage your sites? a. What is the level of effort for SSL, new site setup, database management, and extension management. b. How did you handle datacenter outages. Anyone with Pagoda Box experience, do you like it and did you have problems with Wordpress, Cakephp, Drupal, Expression Engine or Magento? a. Is it expensive for you? b. How has server uptime been? Your direction and comments are greatly appreciated.

    Read the article

  • Why most people change from being a contractor to full time at my companies, but not the other way around?

    - by ????
    I have seen most people changed from being a contractor to being a full time employee, but not the other way around. And that happened in startups that had maybe 20% chance of IPO or being acquired, and another that had maybe a 50% chance. As far as I know, the rate (even for a 3 year experience graphics designer, or a programmer), can be $75 to $80 an hour. While a programmer with 15 years of experience may get $120,000 per year. So, the programmer with 15 years of programming experience earns $10,000 per month. At the same time, the programmer with 3 year of experience or the graphics designer will get $14,000 per month ($80 x 22 days x 8 hours). I know I have to buy my own insurance, but I can't imagine buying those for $4,000 each month... maybe $200, $300 at most. I probably need to pay Social Security (FICA) both way (myself and as self-employed = $210 x 2), but still, each month there will be extra $3,000 of income. So is the above calculation correct? But most often, I do see contractors wanting or becoming full time, but not full time employees becoming a contractor. Does somebody know what the reason is?

    Read the article

  • How to find and fix performance problems in ORM powered applications

    - by FransBouma
    Once in a while we get requests about how to fix performance problems with our framework. As it comes down to following the same steps and looking into the same things every single time, I decided to write a blogpost about it instead, so more people can learn from this and solve performance problems in their O/R mapper powered applications. In some parts it's focused on LLBLGen Pro but it's also usable for other O/R mapping frameworks, as the vast majority of performance problems in O/R mapper powered applications are not specific for a certain O/R mapper framework. Too often, the developer looks at the wrong part of the application, trying to fix what isn't a problem in that part, and getting frustrated that 'things are so slow with <insert your favorite framework X here>'. I'm in the O/R mapper business for a long time now (almost 10 years, full time) and as it's a small world, we O/R mapper developers know almost all tricks to pull off by now: we all know what to do to make task ABC faster and what compromises (because there are almost always compromises) to deal with if we decide to make ABC faster that way. Some O/R mapper frameworks are faster in X, others in Y, but you can be sure the difference is mainly a result of a compromise some developers are willing to deal with and others aren't. That's why the O/R mapper frameworks on the market today are different in many ways, even though they all fetch and save entities from and to a database. I'm not suggesting there's no room for improvement in today's O/R mapper frameworks, there always is, but it's not a matter of 'the slowness of the application is caused by the O/R mapper' anymore. Perhaps query generation can be optimized a bit here, row materialization can be optimized a bit there, but it's mainly coming down to milliseconds. Still worth it if you're a framework developer, but it's not much compared to the time spend inside databases and in user code: if a complete fetch takes 40ms or 50ms (from call to entity object collection), it won't make a difference for your application as that 10ms difference won't be noticed. That's why it's very important to find the real locations of the problems so developers can fix them properly and don't get frustrated because their quest to get a fast, performing application failed. Performance tuning basics and rules Finding and fixing performance problems in any application is a strict procedure with four prescribed steps: isolate, analyze, interpret and fix, in that order. It's key that you don't skip a step nor make assumptions: these steps help you find the reason of a problem which seems to be there, and how to fix it or leave it as-is. Skipping a step, or when you assume things will be bad/slow without doing analysis will lead to the path of premature optimization and won't actually solve your problems, only create new ones. The most important rule of finding and fixing performance problems in software is that you have to understand what 'performance problem' actually means. Most developers will say "when a piece of software / code is slow, you have a performance problem". But is that actually the case? If I write a Linq query which will aggregate, group and sort 5 million rows from several tables to produce a resultset of 10 rows, it might take more than a couple of milliseconds before that resultset is ready to be consumed by other logic. If I solely look at the Linq query, the code consuming the resultset of the 10 rows and then look at the time it takes to complete the whole procedure, it will appear to me to be slow: all that time taken to produce and consume 10 rows? But if you look closer, if you analyze and interpret the situation, you'll see it does a tremendous amount of work, and in that light it might even be extremely fast. With every performance problem you encounter, always do realize that what you're trying to solve is perhaps not a technical problem at all, but a perception problem. The second most important rule you have to understand is based on the old saying "Penny wise, Pound Foolish": the part which takes e.g. 5% of the total time T for a given task isn't worth optimizing if you have another part which takes a much larger part of the total time T for that same given task. Optimizing parts which are relatively insignificant for the total time taken is not going to bring you better results overall, even if you totally optimize that part away. This is the core reason why analysis of the complete set of application parts which participate in a given task is key to being successful in solving performance problems: No analysis -> no problem -> no solution. One warning up front: hunting for performance will always include making compromises. Fast software can be made maintainable, but if you want to squeeze as much performance out of your software, you will inevitably be faced with the dilemma of compromising one or more from the group {readability, maintainability, features} for the extra performance you think you'll gain. It's then up to you to decide whether it's worth it. In almost all cases it's not. The reason for this is simple: the vast majority of performance problems can be solved by implementing the proper algorithms, the ones with proven Big O-characteristics so you know the performance you'll get plus you know the algorithm will work. The time taken by the algorithm implementing code is inevitable: you already implemented the best algorithm. You might find some optimizations on the technical level but in general these are minor. Let's look at the four steps to see how they guide us through the quest to find and fix performance problems. Isolate The first thing you need to do is to isolate the areas in your application which are assumed to be slow. For example, if your application is a web application and a given page is taking several seconds or even minutes to load, it's a good candidate to check out. It's important to start with the isolate step because it allows you to focus on a single code path per area with a clear begin and end and ignore the rest. The rest of the steps are taken per identified problematic area. Keep in mind that isolation focuses on tasks in an application, not code snippets. A task is something that's started in your application by either another task or the user, or another program, and has a beginning and an end. You can see a task as a piece of functionality offered by your application.  Analyze Once you've determined the problem areas, you have to perform analysis on the code paths of each area, to see where the performance problems occur and which areas are not the problem. This is a multi-layered effort: an application which uses an O/R mapper typically consists of multiple parts: there's likely some kind of interface (web, webservice, windows etc.), a part which controls the interface and business logic, the O/R mapper part and the RDBMS, all connected with either a network or inter-process connections provided by the OS or other means. Each of these parts, including the connectivity plumbing, eat up a part of the total time it takes to complete a task, e.g. load a webpage with all orders of a given customer X. To understand which parts participate in the task / area we're investigating and how much they contribute to the total time taken to complete the task, analysis of each participating task is essential. Start with the code you wrote which starts the task, analyze the code and track the path it follows through your application. What does the code do along the way, verify whether it's correct or not. Analyze whether you have implemented the right algorithms in your code for this particular area. Remember we're looking at one area at a time, which means we're ignoring all other code paths, just the code path of the current problematic area, from begin to end and back. Don't dig in and start optimizing at the code level just yet. We're just analyzing. If your analysis reveals big architectural stupidity, it's perhaps a good idea to rethink the architecture at this point. For the rest, we're analyzing which means we collect data about what could be wrong, for each participating part of the complete application. Reviewing the code you wrote is a good tool to get deeper understanding of what is going on for a given task but ultimately it lacks precision and overview what really happens: humans aren't good code interpreters, computers are. We therefore need to utilize tools to get deeper understanding about which parts contribute how much time to the total task, triggered by which other parts and for example how many times are they called. There are two different kind of tools which are necessary: .NET profilers and O/R mapper / RDBMS profilers. .NET profiling .NET profilers (e.g. dotTrace by JetBrains or Ants by Red Gate software) show exactly which pieces of code are called, how many times they're called, and the time it took to run that piece of code, at the method level and sometimes even at the line level. The .NET profilers are essential tools for understanding whether the time taken to complete a given task / area in your application is consumed by .NET code, where exactly in your code, the path to that code, how many times that code was called by other code and thus reveals where hotspots are located: the areas where a solution can be found. Importantly, they also reveal which areas can be left alone: remember our penny wise pound foolish saying: if a profiler reveals that a group of methods are fast, or don't contribute much to the total time taken for a given task, ignore them. Even if the code in them is perhaps complex and looks like a candidate for optimization: you can work all day on that, it won't matter.  As we're focusing on a single area of the application, it's best to start profiling right before you actually activate the task/area. Most .NET profilers support this by starting the application without starting the profiling procedure just yet. You navigate to the particular part which is slow, start profiling in the profiler, in your application you perform the actions which are considered slow, and afterwards you get a snapshot in the profiler. The snapshot contains the data collected by the profiler during the slow action, so most data is produced by code in the area to investigate. This is important, because it allows you to stay focused on a single area. O/R mapper and RDBMS profiling .NET profilers give you a good insight in the .NET side of things, but not in the RDBMS side of the application. As this article is about O/R mapper powered applications, we're also looking at databases, and the software making it possible to consume the database in your application: the O/R mapper. To understand which parts of the O/R mapper and database participate how much to the total time taken for task T, we need different tools. There are two kind of tools focusing on O/R mappers and database performance profiling: O/R mapper profilers and RDBMS profilers. For O/R mapper profilers, you can look at LLBLGen Prof by hibernating rhinos or the Linq to Sql/LLBLGen Pro profiler by Huagati. Hibernating rhinos also have profilers for other O/R mappers like NHibernate (NHProf) and Entity Framework (EFProf) and work the same as LLBLGen Prof. For RDBMS profilers, you have to look whether the RDBMS vendor has a profiler. For example for SQL Server, the profiler is shipped with SQL Server, for Oracle it's build into the RDBMS, however there are also 3rd party tools. Which tool you're using isn't really important, what's important is that you get insight in which queries are executed during the task / area we're currently focused on and how long they took. Here, the O/R mapper profilers have an advantage as they collect the time it took to execute the query from the application's perspective so they also collect the time it took to transport data across the network. This is important because a query which returns a massive resultset or a resultset with large blob/clob/ntext/image fields takes more time to get transported across the network than a small resultset and a database profiler doesn't take this into account most of the time. Another tool to use in this case, which is more low level and not all O/R mappers support it (though LLBLGen Pro and NHibernate as well do) is tracing: most O/R mappers offer some form of tracing or logging system which you can use to collect the SQL generated and executed and often also other activity behind the scenes. While tracing can produce a tremendous amount of data in some cases, it also gives insight in what's going on. Interpret After we've completed the analysis step it's time to look at the data we've collected. We've done code reviews to see whether we've done anything stupid and which parts actually take place and if the proper algorithms have been implemented. We've done .NET profiling to see which parts are choke points and how much time they contribute to the total time taken to complete the task we're investigating. We've performed O/R mapper profiling and RDBMS profiling to see which queries were executed during the task, how many queries were generated and executed and how long they took to complete, including network transportation. All this data reveals two things: which parts are big contributors to the total time taken and which parts are irrelevant. Both aspects are very important. The parts which are irrelevant (i.e. don't contribute significantly to the total time taken) can be ignored from now on, we won't look at them. The parts which contribute a lot to the total time taken are important to look at. We now have to first look at the .NET profiler results, to see whether the time taken is consumed in our own code, in .NET framework code, in the O/R mapper itself or somewhere else. For example if most of the time is consumed by DbCommand.ExecuteReader, the time it took to complete the task is depending on the time the data is fetched from the database. If there was just 1 query executed, according to tracing or O/R mapper profilers / RDBMS profilers, check whether that query is optimal, uses indexes or has to deal with a lot of data. Interpret means that you follow the path from begin to end through the data collected and determine where, along the path, the most time is contributed. It also means that you have to check whether this was expected or is totally unexpected. My previous example of the 10 row resultset of a query which groups millions of rows will likely reveal that a long time is spend inside the database and almost no time is spend in the .NET code, meaning the RDBMS part contributes the most to the total time taken, the rest is compared to that time, irrelevant. Considering the vastness of the source data set, it's expected this will take some time. However, does it need tweaking? Perhaps all possible tweaks are already in place. In the interpret step you then have to decide that further action in this area is necessary or not, based on what the analysis results show: if the analysis results were unexpected and in the area where the most time is contributed to the total time taken is room for improvement, action should be taken. If not, you can only accept the situation and move on. In all cases, document your decision together with the analysis you've done. If you decide that the perceived performance problem is actually expected due to the nature of the task performed, it's essential that in the future when someone else looks at the application and starts asking questions you can answer them properly and new analysis is only necessary if situations changed. Fix After interpreting the analysis results you've concluded that some areas need adjustment. This is the fix step: you're actively correcting the performance problem with proper action targeted at the real cause. In many cases related to O/R mapper powered applications it means you'll use different features of the O/R mapper to achieve the same goal, or apply optimizations at the RDBMS level. It could also mean you apply caching inside your application (compromise memory consumption over performance) to avoid unnecessary re-querying data and re-consuming the results. After applying a change, it's key you re-do the analysis and interpretation steps: compare the results and expectations with what you had before, to see whether your actions had any effect or whether it moved the problem to a different part of the application. Don't fall into the trap to do partly analysis: do the full analysis again: .NET profiling and O/R mapper / RDBMS profiling. It might very well be that the changes you've made make one part faster but another part significantly slower, in such a way that the overall problem hasn't changed at all. Performance tuning is dealing with compromises and making choices: to use one feature over the other, to accept a higher memory footprint, to go away from the strict-OO path and execute queries directly onto the RDBMS, these are choices and compromises which will cross your path if you want to fix performance problems with respect to O/R mappers or data-access and databases in general. In most cases it's not a big issue: alternatives are often good choices too and the compromises aren't that hard to deal with. What is important is that you document why you made a choice, a compromise: which analysis data, which interpretation led you to the choice made. This is key for good maintainability in the years to come. Most common performance problems with O/R mappers Below is an incomplete list of common performance problems related to data-access / O/R mappers / RDBMS code. It will help you with fixing the hotspots you found in the interpretation step. SELECT N+1: (Lazy-loading specific). Lazy loading triggered performance bottlenecks. Consider a list of Orders bound to a grid. You have a Field mapped onto a related field in Order, Customer.CompanyName. Showing this column in the grid will make the grid fetch (indirectly) for each row the Customer row. This means you'll get for the single list not 1 query (for the orders) but 1+(the number of orders shown) queries. To solve this: use eager loading using a prefetch path to fetch the customers with the orders. SELECT N+1 is easy to spot with an O/R mapper profiler or RDBMS profiler: if you see a lot of identical queries executed at once, you have this problem. Prefetch paths using many path nodes or sorting, or limiting. Eager loading problem. Prefetch paths can help with performance, but as 1 query is fetched per node, it can be the number of data fetched in a child node is bigger than you think. Also consider that data in every node is merged on the client within the parent. This is fast, but it also can take some time if you fetch massive amounts of entities. If you keep fetches small, you can use tuning parameters like the ParameterizedPrefetchPathThreshold setting to get more optimal queries. Deep inheritance hierarchies of type Target Per Entity/Type. If you use inheritance of type Target per Entity / Type (each type in the inheritance hierarchy is mapped onto its own table/view), fetches will join subtype- and supertype tables in many cases, which can lead to a lot of performance problems if the hierarchy has many types. With this problem, keep inheritance to a minimum if possible, or switch to a hierarchy of type Target Per Hierarchy, which means all entities in the inheritance hierarchy are mapped onto the same table/view. Of course this has its own set of drawbacks, but it's a compromise you might want to take. Fetching massive amounts of data by fetching large lists of entities. LLBLGen Pro supports paging (and limiting the # of rows returned), which is often key to process through large sets of data. Use paging on the RDBMS if possible (so a query is executed which returns only the rows in the page requested). When using paging in a web application, be sure that you switch server-side paging on on the datasourcecontrol used. In this case, paging on the grid alone is not enough: this can lead to fetching a lot of data which is then loaded into the grid and paged there. Keep note that analyzing queries for paging could lead to the false assumption that paging doesn't occur, e.g. when the query contains a field of type ntext/image/clob/blob and DISTINCT can't be applied while it should have (e.g. due to a join): the datareader will do DISTINCT filtering on the client. this is a little slower but it does perform paging functionality on the data-reader so it won't fetch all rows even if the query suggests it does. Fetch massive amounts of data because blob/clob/ntext/image fields aren't excluded. LLBLGen Pro supports field exclusion for queries. You can exclude fields (also in prefetch paths) per query to avoid fetching all fields of an entity, e.g. when you don't need them for the logic consuming the resultset. Excluding fields can greatly reduce the amount of time spend on data-transport across the network. Use this optimization if you see that there's a big difference between query execution time on the RDBMS and the time reported by the .NET profiler for the ExecuteReader method call. Doing client-side aggregates/scalar calculations by consuming a lot of data. If possible, try to formulate a scalar query or group by query using the projection system or GetScalar functionality of LLBLGen Pro to do data consumption on the RDBMS server. It's far more efficient to process data on the RDBMS server than to first load it all in memory, then traverse the data in-memory to calculate a value. Using .ToList() constructs inside linq queries. It might be you use .ToList() somewhere in a Linq query which makes the query be run partially in-memory. Example: var q = from c in metaData.Customers.ToList() where c.Country=="Norway" select c; This will actually fetch all customers in-memory and do an in-memory filtering, as the linq query is defined on an IEnumerable<T>, and not on the IQueryable<T>. Linq is nice, but it can often be a bit unclear where some parts of a Linq query might run. Fetching all entities to delete into memory first. To delete a set of entities it's rather inefficient to first fetch them all into memory and then delete them one by one. It's more efficient to execute a DELETE FROM ... WHERE query on the database directly to delete the entities in one go. LLBLGen Pro supports this feature, and so do some other O/R mappers. It's not always possible to do this operation in the context of an O/R mapper however: if an O/R mapper relies on a cache, these kind of operations are likely not supported because they make it impossible to track whether an entity is actually removed from the DB and thus can be removed from the cache. Fetching all entities to update with an expression into memory first. Similar to the previous point: it is more efficient to update a set of entities directly with a single UPDATE query using an expression instead of fetching the entities into memory first and then updating the entities in a loop, and afterwards saving them. It might however be a compromise you don't want to take as it is working around the idea of having an object graph in memory which is manipulated and instead makes the code fully aware there's a RDBMS somewhere. Conclusion Performance tuning is almost always about compromises and making choices. It's also about knowing where to look and how the systems in play behave and should behave. The four steps I provided should help you stay focused on the real problem and lead you towards the solution. Knowing how to optimally use the systems participating in your own code (.NET framework, O/R mapper, RDBMS, network/services) is key for success as well as knowing what's going on inside the application you built. I hope you'll find this guide useful in tracking down performance problems and dealing with them in a useful way.  

    Read the article

  • Question SpeechSynthesizer.SetOutputToAudioStream audio format problem

    - by Chris Kugler
    Hi, I'm currently working on an application which requires transmission of speech encoded to a specific audio format. System.Speech.AudioFormat.SpeechAudioFormatInfo synthFormat = new System.Speech.AudioFormat.SpeechAudioFormatInfo(System.Speech.AudioFormat.EncodingFormat.Pcm, 8000, 16, 1, 16000, 2, null); This states that the audio is in PCM format, 8000 samples per second, 16 bits per sample, mono, 16000 average bytes per second, block alignment of 2. When I attempt to execute the following code there is nothing written to my MemoryStream instance; however when I change from 8000 samples per second up to 11025 the audio data is written successfully. SpeechSynthesizer synthesizer = new SpeechSynthesizer(); waveStream = new MemoryStream(); PromptBuilder pbuilder = new PromptBuilder(); PromptStyle pStyle = new PromptStyle(); pStyle.Emphasis = PromptEmphasis.None; pStyle.Rate = PromptRate.Fast; pStyle.Volume = PromptVolume.ExtraLoud; pbuilder.StartStyle(pStyle); pbuilder.StartParagraph(); pbuilder.StartVoice(VoiceGender.Male, VoiceAge.Teen, 2); pbuilder.StartSentence(); pbuilder.AppendText("This is some text."); pbuilder.EndSentence(); pbuilder.EndVoice(); pbuilder.EndParagraph(); pbuilder.EndStyle(); synthesizer.SetOutputToAudioStream(waveStream, synthFormat); synthesizer.Speak(pbuilder); synthesizer.SetOutputToNull(); There are no exceptions or errors recorded when using a sample rate of 8000 and I couldn't find anything useful in the documentation regarding SetOutputToAudioStream and why it succeeds at 11025 samples per second and not 8000. I have a workaround involving a wav file that I generated and converted to the correct sample rate using some sound editing tools, but I would like to generate the audio from within the application if I can. One particular point of interest was that the SpeechRecognitionEngine accepts that audio format and successfully recognized the speech in my synthesized wave file... Update: Recently discovered that this audio format succeeds for certain installed voices, but fails for others. It fails specifically for LH Michael and LH Michelle, and failure varies for certain voice settings defined in the PromptBuilder.

    Read the article

  • segmented reduction with scattered segments

    - by Christian Rau
    I got to solve a pretty standard problem on the GPU, but I'm quite new to practical GPGPU, so I'm looking for ideas to approach this problem. I have many points in 3-space which are assigned to a very small number of groups (each point belongs to one group), specifically 15 in this case (doesn't ever change). Now I want to compute the mean and covariance matrix of all the groups. So on the CPU it's roughly the same as: for each point p { mean[p.group] += p.pos; covariance[p.group] += p.pos * p.pos; ++count[p.group]; } for each group g { mean[g] /= count[g]; covariance[g] = covariance[g]/count[g] - mean[g]*mean[g]; } Since the number of groups is extremely small, the last step can be done on the CPU (I need those values on the CPU, anyway). The first step is actually just a segmented reduction, but with the segments scattered around. So the first idea I came up with, was to first sort the points by their groups. I thought about a simple bucket sort using atomic_inc to compute bucket sizes and per-point relocation indices (got a better idea for sorting?, atomics may not be the best idea). After that they're sorted by groups and I could possibly come up with an adaption of the segmented scan algorithms presented here. But in this special case, I got a very large amount of data per point (9-10 floats, maybe even doubles if the need arises), so the standard algorithms using a shared memory element per thread and a thread per point might make problems regarding per-multiprocessor resources as shared memory or registers (Ok, much more on compute capability 1.x than 2.x, but still). Due to the very small and constant number of groups I thought there might be better approaches. Maybe there are already existing ideas suited for these specific properties of such a standard problem. Or maybe my general approach isn't that bad and you got ideas for improving the individual steps, like a good sorting algorithm suited for a very small number of keys or some segmented reduction algorithm minimizing shared memory/register usage. I'm looking for general approaches and don't want to use external libraries. FWIW I'm using OpenCL, but it shouldn't really matter as the general concepts of GPU computing don't really differ over the major frameworks.

    Read the article

  • How to read oom-killer syslog messages?

    - by Grant
    I have a Ubuntu 12.04 server which sometimes dies completely - no SSH, no ping, nothing until it is physically rebooted. After the reboot, I see in syslog that the oom-killer killed, well, pretty much everything. There's a lot of detailed memory usage information in them. How do I read these logs to see what caused the OOM issue? The server has far more memory than it needs, so it shouldn't be running out of memory. Oct 25 07:28:04 nldedip4k031 kernel: [87946.529511] oom_kill_process: 9 callbacks suppressed Oct 25 07:28:04 nldedip4k031 kernel: [87946.529514] irqbalance invoked oom-killer: gfp_mask=0x80d0, order=0, oom_adj=0, oom_score_adj=0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529516] irqbalance cpuset=/ mems_allowed=0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529518] Pid: 948, comm: irqbalance Not tainted 3.2.0-55-generic-pae #85-Ubuntu Oct 25 07:28:04 nldedip4k031 kernel: [87946.529519] Call Trace: Oct 25 07:28:04 nldedip4k031 kernel: [87946.529525] [] dump_header.isra.6+0x85/0xc0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529528] [] oom_kill_process+0x5c/0x80 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529530] [] out_of_memory+0xc5/0x1c0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529532] [] __alloc_pages_nodemask+0x72c/0x740 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529535] [] __get_free_pages+0x1c/0x30 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529537] [] get_zeroed_page+0x12/0x20 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529541] [] fill_read_buffer.isra.8+0xaa/0xd0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529543] [] sysfs_read_file+0x7d/0x90 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529546] [] vfs_read+0x8c/0x160 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529548] [] ? fill_read_buffer.isra.8+0xd0/0xd0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529550] [] sys_read+0x3d/0x70 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529554] [] sysenter_do_call+0x12/0x28 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529555] Mem-Info: Oct 25 07:28:04 nldedip4k031 kernel: [87946.529556] DMA per-cpu: Oct 25 07:28:04 nldedip4k031 kernel: [87946.529557] CPU 0: hi: 0, btch: 1 usd: 0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529558] CPU 1: hi: 0, btch: 1 usd: 0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529560] CPU 2: hi: 0, btch: 1 usd: 0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529561] CPU 3: hi: 0, btch: 1 usd: 0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529562] CPU 4: hi: 0, btch: 1 usd: 0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529563] CPU 5: hi: 0, btch: 1 usd: 0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529564] CPU 6: hi: 0, btch: 1 usd: 0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529565] CPU 7: hi: 0, btch: 1 usd: 0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529566] Normal per-cpu: Oct 25 07:28:04 nldedip4k031 kernel: [87946.529567] CPU 0: hi: 186, btch: 31 usd: 179 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529568] CPU 1: hi: 186, btch: 31 usd: 182 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529569] CPU 2: hi: 186, btch: 31 usd: 132 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529570] CPU 3: hi: 186, btch: 31 usd: 175 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529571] CPU 4: hi: 186, btch: 31 usd: 91 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529572] CPU 5: hi: 186, btch: 31 usd: 173 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529573] CPU 6: hi: 186, btch: 31 usd: 159 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529574] CPU 7: hi: 186, btch: 31 usd: 164 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529575] HighMem per-cpu: Oct 25 07:28:04 nldedip4k031 kernel: [87946.529576] CPU 0: hi: 186, btch: 31 usd: 165 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529577] CPU 1: hi: 186, btch: 31 usd: 183 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529578] CPU 2: hi: 186, btch: 31 usd: 185 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529579] CPU 3: hi: 186, btch: 31 usd: 138 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529580] CPU 4: hi: 186, btch: 31 usd: 155 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529581] CPU 5: hi: 186, btch: 31 usd: 104 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529582] CPU 6: hi: 186, btch: 31 usd: 133 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529583] CPU 7: hi: 186, btch: 31 usd: 170 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529586] active_anon:5523 inactive_anon:354 isolated_anon:0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529586] active_file:2815 inactive_file:6849119 isolated_file:0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529587] unevictable:0 dirty:449 writeback:10 unstable:0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529587] free:1304125 slab_reclaimable:104672 slab_unreclaimable:3419 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529588] mapped:2661 shmem:138 pagetables:313 bounce:0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529591] DMA free:4252kB min:780kB low:972kB high:1168kB active_anon:0kB inactive_anon:0kB active_file:4kB inactive_file:0kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15756kB mlocked:0kB dirty:0kB writeback:0kB mapped:0kB shmem:0kB slab_reclaimable:11564kB slab_unreclaimable:4kB kernel_stack:0kB pagetables:0kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:1 all_unreclaimable? yes Oct 25 07:28:04 nldedip4k031 kernel: [87946.529594] lowmem_reserve[]: 0 869 32460 32460 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529599] Normal free:44052kB min:44216kB low:55268kB high:66324kB active_anon:0kB inactive_anon:0kB active_file:616kB inactive_file:568kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:890008kB mlocked:0kB dirty:0kB writeback:0kB mapped:4kB shmem:0kB slab_reclaimable:407124kB slab_unreclaimable:13672kB kernel_stack:992kB pagetables:0kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:2083 all_unreclaimable? yes Oct 25 07:28:04 nldedip4k031 kernel: [87946.529602] lowmem_reserve[]: 0 0 252733 252733 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529606] HighMem free:5168196kB min:512kB low:402312kB high:804112kB active_anon:22092kB inactive_anon:1416kB active_file:10640kB inactive_file:27395920kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:32349872kB mlocked:0kB dirty:1796kB writeback:40kB mapped:10640kB shmem:552kB slab_reclaimable:0kB slab_unreclaimable:0kB kernel_stack:0kB pagetables:1252kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no Oct 25 07:28:04 nldedip4k031 kernel: [87946.529609] lowmem_reserve[]: 0 0 0 0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529611] DMA: 6*4kB 6*8kB 6*16kB 5*32kB 5*64kB 4*128kB 2*256kB 1*512kB 0*1024kB 1*2048kB 0*4096kB = 4232kB Oct 25 07:28:04 nldedip4k031 kernel: [87946.529616] Normal: 297*4kB 180*8kB 119*16kB 73*32kB 67*64kB 47*128kB 35*256kB 13*512kB 5*1024kB 1*2048kB 1*4096kB = 44052kB Oct 25 07:28:04 nldedip4k031 kernel: [87946.529622] HighMem: 1*4kB 6*8kB 27*16kB 11*32kB 2*64kB 1*128kB 0*256kB 0*512kB 4*1024kB 1*2048kB 1260*4096kB = 5168196kB Oct 25 07:28:04 nldedip4k031 kernel: [87946.529627] 6852076 total pagecache pages Oct 25 07:28:04 nldedip4k031 kernel: [87946.529628] 0 pages in swap cache Oct 25 07:28:04 nldedip4k031 kernel: [87946.529629] Swap cache stats: add 0, delete 0, find 0/0 Oct 25 07:28:04 nldedip4k031 kernel: [87946.529630] Free swap = 3998716kB Oct 25 07:28:04 nldedip4k031 kernel: [87946.529631] Total swap = 3998716kB Oct 25 07:28:04 nldedip4k031 kernel: [87946.571914] 8437743 pages RAM Oct 25 07:28:04 nldedip4k031 kernel: [87946.571916] 8209409 pages HighMem Oct 25 07:28:04 nldedip4k031 kernel: [87946.571917] 159556 pages reserved Oct 25 07:28:04 nldedip4k031 kernel: [87946.571917] 6862034 pages shared Oct 25 07:28:04 nldedip4k031 kernel: [87946.571918] 123540 pages non-shared Oct 25 07:28:04 nldedip4k031 kernel: [87946.571919] [ pid ] uid tgid total_vm rss cpu oom_adj oom_score_adj name Oct 25 07:28:04 nldedip4k031 kernel: [87946.571927] [ 421] 0 421 709 152 3 0 0 upstart-udev-br Oct 25 07:28:04 nldedip4k031 kernel: [87946.571929] [ 429] 0 429 773 326 5 -17 -1000 udevd Oct 25 07:28:04 nldedip4k031 kernel: [87946.571931] [ 567] 0 567 772 224 4 -17 -1000 udevd Oct 25 07:28:04 nldedip4k031 kernel: [87946.571932] [ 568] 0 568 772 231 7 -17 -1000 udevd Oct 25 07:28:04 nldedip4k031 kernel: [87946.571934] [ 764] 0 764 712 103 1 0 0 upstart-socket- Oct 25 07:28:04 nldedip4k031 kernel: [87946.571936] [ 772] 103 772 815 164 5 0 0 dbus-daemon Oct 25 07:28:04 nldedip4k031 kernel: [87946.571938] [ 785] 0 785 1671 600 1 -17 -1000 sshd Oct 25 07:28:04 nldedip4k031 kernel: [87946.571940] [ 809] 101 809 7766 380 1 0 0 rsyslogd Oct 25 07:28:04 nldedip4k031 kernel: [87946.571942] [ 869] 0 869 1158 213 3 0 0 getty Oct 25 07:28:04 nldedip4k031 kernel: [87946.571943] [ 873] 0 873 1158 214 6 0 0 getty Oct 25 07:28:04 nldedip4k031 kernel: [87946.571945] [ 911] 0 911 1158 215 3 0 0 getty Oct 25 07:28:04 nldedip4k031 kernel: [87946.571947] [ 912] 0 912 1158 214 2 0 0 getty Oct 25 07:28:04 nldedip4k031 kernel: [87946.571949] [ 914] 0 914 1158 213 1 0 0 getty Oct 25 07:28:04 nldedip4k031 kernel: [87946.571950] [ 916] 0 916 618 86 1 0 0 atd Oct 25 07:28:04 nldedip4k031 kernel: [87946.571952] [ 917] 0 917 655 226 3 0 0 cron Oct 25 07:28:04 nldedip4k031 kernel: [87946.571954] [ 948] 0 948 902 159 3 0 0 irqbalance Oct 25 07:28:04 nldedip4k031 kernel: [87946.571956] [ 993] 0 993 1145 363 3 0 0 master Oct 25 07:28:04 nldedip4k031 kernel: [87946.571957] [ 1002] 104 1002 1162 333 1 0 0 qmgr Oct 25 07:28:04 nldedip4k031 kernel: [87946.571959] [ 1016] 0 1016 730 149 2 0 0 mdadm Oct 25 07:28:04 nldedip4k031 kernel: [87946.571961] [ 1057] 0 1057 6066 2160 3 0 0 /usr/sbin/apach Oct 25 07:28:04 nldedip4k031 kernel: [87946.571963] [ 1086] 0 1086 1158 213 3 0 0 getty Oct 25 07:28:04 nldedip4k031 kernel: [87946.571965] [ 1088] 33 1088 6191 1517 0 0 0 /usr/sbin/apach Oct 25 07:28:04 nldedip4k031 kernel: [87946.571967] [ 1089] 33 1089 6191 1451 1 0 0 /usr/sbin/apach Oct 25 07:28:04 nldedip4k031 kernel: [87946.571969] [ 1090] 33 1090 6175 1451 3 0 0 /usr/sbin/apach Oct 25 07:28:04 nldedip4k031 kernel: [87946.571971] [ 1091] 33 1091 6191 1451 1 0 0 /usr/sbin/apach Oct 25 07:28:04 nldedip4k031 kernel: [87946.571972] [ 1092] 33 1092 6191 1451 0 0 0 /usr/sbin/apach Oct 25 07:28:04 nldedip4k031 kernel: [87946.571974] [ 1109] 33 1109 6191 1517 0 0 0 /usr/sbin/apach Oct 25 07:28:04 nldedip4k031 kernel: [87946.571976] [ 1151] 33 1151 6191 1451 1 0 0 /usr/sbin/apach Oct 25 07:28:04 nldedip4k031 kernel: [87946.571978] [ 1201] 104 1201 1803 652 1 0 0 tlsmgr Oct 25 07:28:04 nldedip4k031 kernel: [87946.571980] [ 2475] 0 2475 2435 812 0 0 0 sshd Oct 25 07:28:04 nldedip4k031 kernel: [87946.571982] [ 2494] 0 2494 1745 839 1 0 0 bash Oct 25 07:28:04 nldedip4k031 kernel: [87946.571984] [ 2573] 0 2573 3394 1689 0 0 0 sshd Oct 25 07:28:04 nldedip4k031 kernel: [87946.571986] [ 2589] 0 2589 5014 457 3 0 0 rsync Oct 25 07:28:04 nldedip4k031 kernel: [87946.571988] [ 2590] 0 2590 7970 522 1 0 0 rsync Oct 25 07:28:04 nldedip4k031 kernel: [87946.571990] [ 2652] 104 2652 1150 326 5 0 0 pickup Oct 25 07:28:04 nldedip4k031 kernel: [87946.571992] Out of memory: Kill process 421 (upstart-udev-br) score 1 or sacrifice child Oct 25 07:28:04 nldedip4k031 kernel: [87946.572407] Killed process 421 (upstart-udev-br) total-vm:2836kB, anon-rss:156kB, file-rss:452kB Oct 25 07:28:04 nldedip4k031 kernel: [87946.573107] init: upstart-udev-bridge main process (421) killed by KILL signal Oct 25 07:28:04 nldedip4k031 kernel: [87946.573126] init: upstart-udev-bridge main process ended, respawning Oct 25 07:28:34 nldedip4k031 kernel: [87976.461570] irqbalance invoked oom-killer: gfp_mask=0x80d0, order=0, oom_adj=0, oom_score_adj=0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461573] irqbalance cpuset=/ mems_allowed=0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461576] Pid: 948, comm: irqbalance Not tainted 3.2.0-55-generic-pae #85-Ubuntu Oct 25 07:28:34 nldedip4k031 kernel: [87976.461578] Call Trace: Oct 25 07:28:34 nldedip4k031 kernel: [87976.461585] [] dump_header.isra.6+0x85/0xc0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461588] [] oom_kill_process+0x5c/0x80 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461591] [] out_of_memory+0xc5/0x1c0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461595] [] __alloc_pages_nodemask+0x72c/0x740 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461599] [] __get_free_pages+0x1c/0x30 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461602] [] get_zeroed_page+0x12/0x20 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461606] [] fill_read_buffer.isra.8+0xaa/0xd0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461609] [] sysfs_read_file+0x7d/0x90 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461613] [] vfs_read+0x8c/0x160 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461616] [] ? fill_read_buffer.isra.8+0xd0/0xd0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461619] [] sys_read+0x3d/0x70 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461624] [] sysenter_do_call+0x12/0x28 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461626] Mem-Info: Oct 25 07:28:34 nldedip4k031 kernel: [87976.461628] DMA per-cpu: Oct 25 07:28:34 nldedip4k031 kernel: [87976.461629] CPU 0: hi: 0, btch: 1 usd: 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461631] CPU 1: hi: 0, btch: 1 usd: 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461633] CPU 2: hi: 0, btch: 1 usd: 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461634] CPU 3: hi: 0, btch: 1 usd: 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461636] CPU 4: hi: 0, btch: 1 usd: 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461638] CPU 5: hi: 0, btch: 1 usd: 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461639] CPU 6: hi: 0, btch: 1 usd: 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461641] CPU 7: hi: 0, btch: 1 usd: 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461642] Normal per-cpu: Oct 25 07:28:34 nldedip4k031 kernel: [87976.461644] CPU 0: hi: 186, btch: 31 usd: 61 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461646] CPU 1: hi: 186, btch: 31 usd: 49 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461647] CPU 2: hi: 186, btch: 31 usd: 8 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461649] CPU 3: hi: 186, btch: 31 usd: 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461651] CPU 4: hi: 186, btch: 31 usd: 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461652] CPU 5: hi: 186, btch: 31 usd: 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461654] CPU 6: hi: 186, btch: 31 usd: 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461656] CPU 7: hi: 186, btch: 31 usd: 30 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461657] HighMem per-cpu: Oct 25 07:28:34 nldedip4k031 kernel: [87976.461658] CPU 0: hi: 186, btch: 31 usd: 4 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461660] CPU 1: hi: 186, btch: 31 usd: 204 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461662] CPU 2: hi: 186, btch: 31 usd: 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461663] CPU 3: hi: 186, btch: 31 usd: 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461665] CPU 4: hi: 186, btch: 31 usd: 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461667] CPU 5: hi: 186, btch: 31 usd: 31 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461668] CPU 6: hi: 186, btch: 31 usd: 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461670] CPU 7: hi: 186, btch: 31 usd: 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461674] active_anon:5441 inactive_anon:412 isolated_anon:0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461674] active_file:2668 inactive_file:6922842 isolated_file:0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461675] unevictable:0 dirty:836 writeback:0 unstable:0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461676] free:1231664 slab_reclaimable:105781 slab_unreclaimable:3399 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461677] mapped:2649 shmem:138 pagetables:313 bounce:0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461682] DMA free:4248kB min:780kB low:972kB high:1168kB active_anon:0kB inactive_anon:0kB active_file:0kB inactive_file:4kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:15756kB mlocked:0kB dirty:0kB writeback:0kB mapped:0kB shmem:0kB slab_reclaimable:11560kB slab_unreclaimable:4kB kernel_stack:0kB pagetables:0kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:5687 all_unreclaimable? yes Oct 25 07:28:34 nldedip4k031 kernel: [87976.461686] lowmem_reserve[]: 0 869 32460 32460 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461693] Normal free:44184kB min:44216kB low:55268kB high:66324kB active_anon:0kB inactive_anon:0kB active_file:20kB inactive_file:1096kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:890008kB mlocked:0kB dirty:4kB writeback:0kB mapped:4kB shmem:0kB slab_reclaimable:411564kB slab_unreclaimable:13592kB kernel_stack:992kB pagetables:0kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:1816 all_unreclaimable? yes Oct 25 07:28:34 nldedip4k031 kernel: [87976.461697] lowmem_reserve[]: 0 0 252733 252733 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461703] HighMem free:4878224kB min:512kB low:402312kB high:804112kB active_anon:21764kB inactive_anon:1648kB active_file:10652kB inactive_file:27690268kB unevictable:0kB isolated(anon):0kB isolated(file):0kB present:32349872kB mlocked:0kB dirty:3340kB writeback:0kB mapped:10592kB shmem:552kB slab_reclaimable:0kB slab_unreclaimable:0kB kernel_stack:0kB pagetables:1252kB unstable:0kB bounce:0kB writeback_tmp:0kB pages_scanned:0 all_unreclaimable? no Oct 25 07:28:34 nldedip4k031 kernel: [87976.461708] lowmem_reserve[]: 0 0 0 0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461711] DMA: 8*4kB 7*8kB 6*16kB 5*32kB 5*64kB 4*128kB 2*256kB 1*512kB 0*1024kB 1*2048kB 0*4096kB = 4248kB Oct 25 07:28:34 nldedip4k031 kernel: [87976.461719] Normal: 272*4kB 178*8kB 76*16kB 52*32kB 42*64kB 36*128kB 23*256kB 20*512kB 7*1024kB 2*2048kB 1*4096kB = 44176kB Oct 25 07:28:34 nldedip4k031 kernel: [87976.461727] HighMem: 1*4kB 45*8kB 31*16kB 24*32kB 5*64kB 3*128kB 1*256kB 2*512kB 4*1024kB 2*2048kB 1188*4096kB = 4877852kB Oct 25 07:28:34 nldedip4k031 kernel: [87976.461736] 6925679 total pagecache pages Oct 25 07:28:34 nldedip4k031 kernel: [87976.461737] 0 pages in swap cache Oct 25 07:28:34 nldedip4k031 kernel: [87976.461739] Swap cache stats: add 0, delete 0, find 0/0 Oct 25 07:28:34 nldedip4k031 kernel: [87976.461740] Free swap = 3998716kB Oct 25 07:28:34 nldedip4k031 kernel: [87976.461741] Total swap = 3998716kB Oct 25 07:28:34 nldedip4k031 kernel: [87976.524951] 8437743 pages RAM Oct 25 07:28:34 nldedip4k031 kernel: [87976.524953] 8209409 pages HighMem Oct 25 07:28:34 nldedip4k031 kernel: [87976.524954] 159556 pages reserved Oct 25 07:28:34 nldedip4k031 kernel: [87976.524955] 6936141 pages shared Oct 25 07:28:34 nldedip4k031 kernel: [87976.524956] 124602 pages non-shared Oct 25 07:28:34 nldedip4k031 kernel: [87976.524957] [ pid ] uid tgid total_vm rss cpu oom_adj oom_score_adj name Oct 25 07:28:34 nldedip4k031 kernel: [87976.524966] [ 429] 0 429 773 326 5 -17 -1000 udevd Oct 25 07:28:34 nldedip4k031 kernel: [87976.524968] [ 567] 0 567 772 224 4 -17 -1000 udevd Oct 25 07:28:34 nldedip4k031 kernel: [87976.524971] [ 568] 0 568 772 231 7 -17 -1000 udevd Oct 25 07:28:34 nldedip4k031 kernel: [87976.524973] [ 764] 0 764 712 103 3 0 0 upstart-socket- Oct 25 07:28:34 nldedip4k031 kernel: [87976.524976] [ 772] 103 772 815 164 2 0 0 dbus-daemon Oct 25 07:28:34 nldedip4k031 kernel: [87976.524979] [ 785] 0 785 1671 600 1 -17 -1000 sshd Oct 25 07:28:34 nldedip4k031 kernel: [87976.524981] [ 809] 101 809 7766 380 1 0 0 rsyslogd Oct 25 07:28:34 nldedip4k031 kernel: [87976.524983] [ 869] 0 869 1158 213 3 0 0 getty Oct 25 07:28:34 nldedip4k031 kernel: [87976.524986] [ 873] 0 873 1158 214 6 0 0 getty Oct 25 07:28:34 nldedip4k031 kernel: [87976.524988] [ 911] 0 911 1158 215 3 0 0 getty Oct 25 07:28:34 nldedip4k031 kernel: [87976.524990] [ 912] 0 912 1158 214 2 0 0 getty Oct 25 07:28:34 nldedip4k031 kernel: [87976.524992] [ 914] 0 914 1158 213 1 0 0 getty Oct 25 07:28:34 nldedip4k031 kernel: [87976.524995] [ 916] 0 916 618 86 1 0 0 atd Oct 25 07:28:34 nldedip4k031 kernel: [87976.524997] [ 917] 0 917 655 226 3 0 0 cron Oct 25 07:28:34 nldedip4k031 kernel: [87976.524999] [ 948] 0 948 902 159 5 0 0 irqbalance Oct 25 07:28:34 nldedip4k031 kernel: [87976.525002] [ 993] 0 993 1145 363 3 0 0 master Oct 25 07:28:34 nldedip4k031 kernel: [87976.525004] [ 1002] 104 1002 1162 333 1 0 0 qmgr Oct 25 07:28:34 nldedip4k031 kernel: [87976.525007] [ 1016] 0 1016 730 149 2 0 0 mdadm Oct 25 07:28:34 nldedip4k031 kernel: [87976.525009] [ 1057] 0 1057 6066 2160 3 0 0 /usr/sbin/apach Oct 25 07:28:34 nldedip4k031 kernel: [87976.525012] [ 1086] 0 1086 1158 213 3 0 0 getty Oct 25 07:28:34 nldedip4k031 kernel: [87976.525014] [ 1088] 33 1088 6191 1517 0 0 0 /usr/sbin/apach Oct 25 07:28:34 nldedip4k031 kernel: [87976.525017] [ 1089] 33 1089 6191 1451 1 0 0 /usr/sbin/apach Oct 25 07:28:34 nldedip4k031 kernel: [87976.525019] [ 1090] 33 1090 6175 1451 1 0 0 /usr/sbin/apach Oct 25 07:28:34 nldedip4k031 kernel: [87976.525021] [ 1091] 33 1091 6191 1451 1 0 0 /usr/sbin/apach Oct 25 07:28:34 nldedip4k031 kernel: [87976.525024] [ 1092] 33 1092 6191 1451 0 0 0 /usr/sbin/apach Oct 25 07:28:34 nldedip4k031 kernel: [87976.525026] [ 1109] 33 1109 6191 1517 0 0 0 /usr/sbin/apach Oct 25 07:28:34 nldedip4k031 kernel: [87976.525029] [ 1151] 33 1151 6191 1451 1 0 0 /usr/sbin/apach Oct 25 07:28:34 nldedip4k031 kernel: [87976.525031] [ 1201] 104 1201 1803 652 1 0 0 tlsmgr Oct 25 07:28:34 nldedip4k031 kernel: [87976.525033] [ 2475] 0 2475 2435 812 0 0 0 sshd Oct 25 07:28:34 nldedip4k031 kernel: [87976.525036] [ 2494] 0 2494 1745 839 1 0 0 bash Oct 25 07:28:34 nldedip4k031 kernel: [87976.525038] [ 2573] 0 2573 3394 1689 3 0 0 sshd Oct 25 07:28:34 nldedip4k031 kernel: [87976.525040] [ 2589] 0 2589 5014 457 3 0 0 rsync Oct 25 07:28:34 nldedip4k031 kernel: [87976.525043] [ 2590] 0 2590 7970 522 1 0 0 rsync Oct 25 07:28:34 nldedip4k031 kernel: [87976.525045] [ 2652] 104 2652 1150 326 5 0 0 pickup Oct 25 07:28:34 nldedip4k031 kernel: [87976.525048] [ 2847] 0 2847 709 89 0 0 0 upstart-udev-br Oct 25 07:28:34 nldedip4k031 kernel: [87976.525050] Out of memory: Kill process 764 (upstart-socket-) score 1 or sacrifice child Oct 25 07:28:34 nldedip4k031 kernel: [87976.525484] Killed process 764 (upstart-socket-) total-vm:2848kB, anon-rss:204kB, file-rss:208kB Oct 25 07:28:34 nldedip4k031 kernel: [87976.526161] init: upstart-socket-bridge main process (764) killed by KILL signal Oct 25 07:28:34 nldedip4k031 kernel: [87976.526180] init: upstart-socket-bridge main process ended, respawning Oct 25 07:28:44 nldedip4k031 kernel: [87986.439671] irqbalance invoked oom-killer: gfp_mask=0x80d0, order=0, oom_adj=0, oom_score_adj=0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439674] irqbalance cpuset=/ mems_allowed=0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439676] Pid: 948, comm: irqbalance Not tainted 3.2.0-55-generic-pae #85-Ubuntu Oct 25 07:28:44 nldedip4k031 kernel: [87986.439678] Call Trace: Oct 25 07:28:44 nldedip4k031 kernel: [87986.439684] [] dump_header.isra.6+0x85/0xc0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439686] [] oom_kill_process+0x5c/0x80 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439688] [] out_of_memory+0xc5/0x1c0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439691] [] __alloc_pages_nodemask+0x72c/0x740 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439694] [] __get_free_pages+0x1c/0x30 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439696] [] get_zeroed_page+0x12/0x20 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439699] [] fill_read_buffer.isra.8+0xaa/0xd0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439702] [] sysfs_read_file+0x7d/0x90 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439704] [] vfs_read+0x8c/0x160 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439707] [] ? fill_read_buffer.isra.8+0xd0/0xd0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439709] [] sys_read+0x3d/0x70 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439712] [] sysenter_do_call+0x12/0x28 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439714] Mem-Info: Oct 25 07:28:44 nldedip4k031 kernel: [87986.439714] DMA per-cpu: Oct 25 07:28:44 nldedip4k031 kernel: [87986.439716] CPU 0: hi: 0, btch: 1 usd: 0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439717] CPU 1: hi: 0, btch: 1 usd: 0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439718] CPU 2: hi: 0, btch: 1 usd: 0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439719] CPU 3: hi: 0, btch: 1 usd: 0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439720] CPU 4: hi: 0, btch: 1 usd: 0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439721] CPU 5: hi: 0, btch: 1 usd: 0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439722] CPU 6: hi: 0, btch: 1 usd: 0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439723] CPU 7: hi: 0, btch: 1 usd: 0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439724] Normal per-cpu: Oct 25 07:28:44 nldedip4k031 kernel: [87986.439725] CPU 0: hi: 186, btch: 31 usd: 0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439726] CPU 1: hi: 186, btch: 31 usd: 0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439727] CPU 2: hi: 186, btch: 31 usd: 0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439728] CPU 3: hi: 186, btch: 31 usd: 0 Oct 25 07:28:44 nldedip4k031 kernel: [87986.439729] CPU 4: hi: 186, btch: 31 usd: 0 Oct 25 07:33:48 nldedip4k031 kernel: imklog 5.8.6, log source = /proc/kmsg started. Oct 25 07:33:48 nldedip4k031 rsyslogd: [origin software="rsyslogd" swVersion="5.8.6" x-pid="2880" x-info="http://www.rsyslog.com"] start Oct 25 07:33:48 nldedip4k031 rsyslogd: rsyslogd's groupid changed to 103 Oct 25 07:33:48 nldedip4k031 rsyslogd: rsyslogd's userid changed to 101 Oct 25 07:33:48 nldedip4k031 rsyslogd-2039: Could not open output pipe '/dev/xconsole' [try http://www.rsyslog.com/e/2039 ]

    Read the article

  • Why do we get a sudden spike in response times?

    - by Christian Hagelid
    We have an API that is implemented using ServiceStack which is hosted in IIS. While performing load testing of the API we discovered that the response times are good but that they deteriorate rapidly as soon as we hit about 3,500 concurrent users per server. We have two servers and when hitting them with 7,000 users the average response times sit below 500ms for all endpoints. The boxes are behind a load balancer so we get 3,500 concurrents per server. However as soon as we increase the number of total concurrent users we see a significant increase in response times. Increasing the concurrent users to 5,000 per server gives us an average response time per endpoint of around 7 seconds. The memory and CPU on the servers are quite low, both while the response times are good and when after they deteriorate. At peak with 10,000 concurrent users the CPU averages just below 50% and the RAM sits around 3-4 GB out of 16. This leaves us thinking that we are hitting some kind of limit somewhere. The below screenshot shows some key counters in perfmon during a load test with a total of 10,000 concurrent users. The highlighted counter is requests/second. To the right of the screenshot you can see the requests per second graph becoming really erratic. This is the main indicator for slow response times. As soon as we see this pattern we notice slow response times in the load test. How do we go about troubleshooting this performance issue? We are trying to identify if this is a coding issue or a configuration issue. Are there any settings in web.config or IIS that could explain this behaviour? The application pool is running .NET v4.0 and the IIS version is 7.5. The only change we have made from the default settings is to update the application pool Queue Length value from 1,000 to 5,000. We have also added the following config settings to the Aspnet.config file: <system.web> <applicationPool maxConcurrentRequestsPerCPU="5000" maxConcurrentThreadsPerCPU="0" requestQueueLimit="5000" /> </system.web> More details: The purpose of the API is to combine data from various external sources and return as JSON. It is currently using an InMemory cache implementation to cache individual external calls at the data layer. The first request to a resource will fetch all data required and any subsequent requests for the same resource will get results from the cache. We have a 'cache runner' that is implemented as a background process that updates the information in the cache at certain set intervals. We have added locking around the code that fetches data from the external resources. We have also implemented the services to fetch the data from the external sources in an asynchronous fashion so that the endpoint should only be as slow as the slowest external call (unless we have data in the cache of course). This is done using the System.Threading.Tasks.Task class. Could we be hitting a limitation in terms of number of threads available to the process?

    Read the article

  • Does this prove a network bandwidth bottleneck?

    - by Yuji Tomita
    I've incorrectly assumed that my internal AB testing means my server can handle 1k concurrency @3k hits per second. My theory at at the moment is that the network is the bottleneck. The server can't send enough data fast enough. External testing from blitz.io at 1k concurrency shows my hits/s capping off at 180, with pages taking longer and longer to respond as the server is only able to return 180 per second. I've served a blank file from nginx and benched it: it scales 1:1 with concurrency. Now to rule out IO / memcached bottlenecks (nginx normally pulls from memcached), I serve up a static version of the cached page from the filesystem. The results are very similar to my original test; I'm capped at around 180 RPS. Splitting the HTML page in half gives me double the RPS, so it's definitely limited by the size of the page. If I internally ApacheBench from the local server, I get consistent results of around 4k RPS on both the Full Page and the Half Page, at high transfer rates. Transfer rate: 62586.14 [Kbytes/sec] received If I AB from an external server, I get around 180RPS - same as the blitz.io results. How do I know it's not intentional throttling? If I benchmark from multiple external servers, all results become poor which leads me to believe the problem is in MY servers outbound traffic, not a download speed issue with my benchmarking servers / blitz.io. So I'm back to my conclusion that my server can't send data fast enough. Am I right? Are there other ways to interpret this data? Is the solution/optimization to set up multiple servers + load balancing that can each serve 180 hits per second? I'm quite new to server optimization, so I'd appreciate any confirmation interpreting this data. Outbound traffic Here's more information about the outbound bandwidth: The network graph shows a maximum output of 16 Mb/s: 16 megabits per second. Doesn't sound like much at all. Due to a suggestion about throttling, I looked into this and found that linode has a 50mbps cap (which I'm not even close to hitting, apparently). I had it raised to 100mbps. Since linode caps my traffic, and I'm not even hitting it, does this mean that my server should indeed be capable of outputting up to 100mbps but is limited by some other internal bottleneck? I just don't understand how networks at this large of a scale work; can they literally send data as fast as they can read from the HDD? Is the network pipe that big? In conclusion 1: Based on the above, I'm thinking I can definitely raise my 180RPS by adding an nginx load balancer on top of a multi nginx server setup at exactly 180RPS per server behind the LB. 2: If linode has a 50/100mbit limit that I'm not hitting at all, there must be something I can do to hit that limit with my single server setup. If I can read / transmit data fast enough locally, and linode even bothers to have a 50mbit/100mbit cap, there must be an internal bottleneck that's not allowing me to hit those caps that I'm not sure how to detect. Correct? I realize the question is huge and vague now, but I'm not sure how to condense it. Any input is appreciated on any conclusion I've made.

    Read the article

< Previous Page | 35 36 37 38 39 40 41 42 43 44 45 46  | Next Page >