Search Results

Search found 29837 results on 1194 pages for 'number to word'.

Page 264/1194 | < Previous Page | 260 261 262 263 264 265 266 267 268 269 270 271  | Next Page >

  • Is this Anti-Scraping technique viable with Crawl-Delay?

    - by skibulk
    I want to prevent web scrapers from abusing 1,000,000 on my website. I'd like to do this by returning a "503 Service Unavailable" error code for users that access an abnormal number of pages per minute. I don't want search engine spiders to ever receive the error. My inclination is to set a robots.txt crawl-delay which will ensure spiders access a number of pages per minute under my 503 threshold. Is this an appropriate solution? Do all major search engines support the directive? Could it negatively affect SEO? Are there any other solutions or recommendations?

    Read the article

  • How to solve linear recurrences involving two functions?

    - by Aditya Bahuguna
    Actually I came across a question in Dynamic Programming where we need to find the number of ways to tile a 2 X N area with tiles of given dimensions.. Here is the problem statement Now after a bit of recurrence solving I came out with these. F(n) = F(n-1) + F(n-2) + 2G(n-1), and G(n) = G(n-1) + F(n-1) I know how to solve LR model where one function is there.For large N as is the case in the above problem we can do the matrix exponentiation and achieve O(k^3log(N)) time where k is the minimum number such that for all km F(n) does not depend on F(n-k). The method of solving linear recurrence with matrix exponentiation as it is given in that blog. Now for the LR involving two functions can anyone suggest an approach feasible enough for large N.

    Read the article

  • Design leaderboard ratings for quiz games

    - by PeterK
    Back in March 2011 i started the following post: How to design a leaderboard? Now my quiz game have been out for approximately a year and sold pretty decently. I am working on to update the game design and is again looking into the leaderboard design to make it better as i am not happy with it. Currently i rate players on number of correct answers, which is not good as it does not consider things like number of games, difficulty levels etc. I also have "extended" stats behind the UITableView (Leaderboard). A player can play based on three levels of difficulty: hard, medium or easy Difficulty levels can be mixed between players in a game Each game can be one to six players, so there can be single games or duels Between 2 and 30 questions per game As i am considering integrating Game Center Leaderboard i need to design a better rating system so i would like to ask for some ideas how to do the rating based on the above. I am thinking about how much a point would be worth and what it includes.

    Read the article

  • Using The Windows Server AppFabric Cache with ASP.NET

    Did you know that you can use the AppFabric Cache with ASP.NET?  AppFabric Cache provides an ASP.NET session state provider.  There are a number of reasons that you would want to consider using the AppFabric Cache instead of other caching technologies, including the built in ASP.NET caching. The AppFabric Cache provides a number of benefits to ASP.NET programmers.  When web applications need to maintain state, especially across a Web Farm, or needs to maintain objects across restarts...Did you know that DotNetSlackers also publishes .net articles written by top known .net Authors? We already have over 80 articles in several categories including Silverlight. Take a look: here.

    Read the article

  • Is there any difference between processor and core?

    - by Salvador
    The following two command seems to give me different information about the same hardware srs@ubuntu:~$ cat /proc/cpuinfo | grep -e processor -e cores processor : 0 cpu cores : 4 processor : 1 cpu cores : 4 processor : 2 cpu cores : 4 processor : 3 cpu cores : 4 srs@ubuntu:~$ sudo dmidecode -t processor # dmidecode 2.9 SMBIOS 2.6 present. Handle 0x0004, DMI type 4, 42 bytes Processor Information Socket Designation: LGA1155 Type: Central Processor Family: <OUT OF SPEC> Manufacturer: Intel ID: A7 06 02 00 FF FB EB BF Version: Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz Voltage: 1.0 V External Clock: 100 MHz Max Speed: 3800 MHz Current Speed: 3300 MHz Status: Populated, Enabled Upgrade: Other L1 Cache Handle: 0x0005 L2 Cache Handle: 0x0006 L3 Cache Handle: 0x0007 Serial Number: To Be Filled By O.E.M. Asset Tag: To Be Filled By O.E.M. Part Number: To Be Filled By O.E.M. Core Count: 4 Core Enabled: 1 Characteristics: 64-bit capable Until today I thought I had a single processor with 4 independent cores. I also thought that within each core can be used different threads.

    Read the article

  • How to reboot into Windows from Ubuntu?

    - by andrewsomething
    I'm looking for a way to reboot into Windows from Ubuntu on a 10.10/Vista dual boot system. The specific use case is that I would like to be able to ssh into my running Ubuntu instance and issue a command that will initiate a reboot directly into Windows. I found a promising blog post, but the script that it suggests isn't working: #!/bin/bash WINDOWS_ENTRY=`grep menuentry /boot/grub/grub.cfg | grep --line-number Windows` MENU_NUMBER=$(( `echo $WINDOWS_ENTRY | sed -e "s/:.*//"` - 1 )) sudo grub-reboot $MENU_NUMBER sudo reboot man grub-reboot isn't much help, but it seems to be leading me in the right direction: set the default boot entry for GRUB, for the next boot only WINDOWS_ENTRY=`grep menuentry /boot/grub/grub.cfg | grep --line-number Windows` MENU_NUMBER=$(( `echo $WINDOWS_ENTRY | sed -e "s/:.*//"` - 1 )) echo $MENU_NUMBER This returns the expected value, but on reboot the first menu entry is still highlighted. Any ideas why this isn't working or suggestions for other solutions?

    Read the article

  • Best Practices for High Volume CPA Import Operations with ebXML in B2B 11g

    - by Shub Lahiri, A-Team
    Background B2B 11g supports ebXML messaging protocol, where multiple CPAs can be imported via command-line utilities.  This note highlights one aspect of the best practices for import of CPA, when large numbers of CPAs in the excess of several hundreds are required to be maintained within the B2B repository. Symptoms The import of CPA usually is a 2-step process, namely creating a soa.zip file using b2bcpaimport utility based on a CPA properties file and then using b2bimport to import the b2b repository.  The commands are provided below: ant -f ant-b2b-util.xml b2bcpaimport -Dpropfile="<Path to cpp_cpa.properties>" -Dstandard=true ant -f ant-b2b-util.xml b2bimport -Dlocalfile=true -Dexportfile="<Path to soa.zip>" -Doverwrite=true Usually the first command completes fairly quickly regardless of the number of CPAs in the repository. However, as the number of trading partners within the repository goes up, the time to complete the second command could go up to ~30 secs per operation. So, this could add up to a significant amount, if there is a need to import hundreds of CPA in a production system within a limited downtime, maintenance window.  Remedy In situations, where there is a large number of entries to be imported, it is best to setup a staging environment and go through the import operation of each individual CPA in an empty repository. Since, this will be done in an empty repository, the time taken for completion should be reasonable.  After all the partner profiles have been imported, a full repository export can be taken to capture the metadata for all the entries in one file.  If this single file with all the partner entries is imported in a loaded repository, the total time taken for import of all the CPAs should see a dramatic reduction. Results Let us take a look at the numbers to see the benefit of this approach. With a pre-loaded repository of ~400 partners, the individual import time for each entry takes ~30 secs. So, if we had to import another 100 partners, the individual entries will take ~50 minutes (100 times ~30 secs). On the other hand, if we prepare the repository export file of the same 100 partners from a staging environment earlier, the import takes about ~5 mins. The total processing time for the loading of metadata, specially in a production environment, can thus be shortened by almost a factor of 10. Summary The following diagram summarizes the entire approach and process. Acknowledgements The material posted here has been compiled with the help from B2B Engineering and Product Management teams.

    Read the article

  • Debugging Minimum Translation Vector

    - by SyntheCypher
    I implemented the minimum translation vector from codezealot's tutorial on SAT (Separating Axis Theorem) but I'm having an issue I can't quite figure out. Here's the example I have: As you can see in top and bottom left images regardless of the side the of the green car which red car is penetrating the MTV for the red car still remains as a negative number also here is the same example when the front of the red car is facing the opposite direction the number will always be positive. When the red car is past the half way through the green car it should switch polarity. I thought I'd compensated for this in my code, but apparently not either that or it's a bug I can find. Here is my function for finding and returning the MTV, any help would be much appreciated: Code

    Read the article

  • Port numbers appended to anchor tags

    - by glifchits
    I've built a static site. Locally, when I serve the content with python -m SimpleHTTPServer everything behaves normally, but when I copy the HTML onto the server and browse the site at the server's URL, some links will have a port number appended to the domain. For example: url.com:84/path where the correct path is url.com/path. The port number is usually different, always between 81-85. It is an Apache server. I'm not experienced with web server configuration, and I'm not the admin of the server. Let me know if there is more information that can help solve my problem. ~> cat /etc/*release* SuSE SLES-8 (i386) VERSION = 8.1 UnitedLinux 1.0 (i586) VERSION = 1.0 LSB_VERSION="1.2" DISTRIB_ID="UnitedLinux" DISTRIB_RELEASE="1.0" DISTRIB_DESCRIPTION="UnitedLinux 1.0 (i586)"

    Read the article

  • Why is this 8 puzzle unsolvable?

    - by Ashwin
    I am developing a 8 puzzle game. I went through the rules in this (see Detecting Unsolvable Puzzles) link, which tell you how to detect if an initial state is unsolvable. It says that if the number of inversions is odd, then the goal state cannot be reached and if even the goal state can be reached. Inversion is defined as Given a board, an inversion is any pair of blocks i and j where i < j but i appears after j when considering the board in row-major order (row 0, followed by row 1, and so forth). There is a 8-puzzle solver(applet) here. Choose 8-puzzle from the options. 1,0,3,2,4,5,6,7,8 and 7,0,2,8,5,3,6,4,1 As you can see both of them contain an even number of inversions. Still the program says that the puzzle is unsolvable. So is the Princeton link wrong?

    Read the article

  • How You Helped Shape Java EE 7...

    - by reza_rahman
    For many of us working with the JCP for years, the commitment to transparency and openness is very clear. For others, perhaps the most visible sign to date of this high regard for grassroots level input is a survey on Java EE 7 gathered a few months ago. The survey was designed to get open feedback on a number of critical issues central to the Java EE 7 umbrella specification including what APIs to include in the standard. The survey was highly successful with a large number of high quality responses. With Java EE 7 under our belt and the horizons for Java EE 8 emerging, this is a good time to thank everyone that took the survey once again for their thoughts and let you know what the impact of your voice actually was. I've posted the details on my personal blog. I hope you are encouraged by how your input to the survey helped shape Java EE 7 and continues to shape Java EE 8. Maybe now is the time for you to get more involved :-)?

    Read the article

  • ATG Live Webcast Dec. 6th: Minimizing EBS Maintenance Downtimes

    - by Bill Sawyer
    This webcast provides an overview of the plans and decisions you can make, and the actions you can take, that will help you minimize maintenance downtimes for your E-Business Suite instances. It is targeted to system administrators, DBAs, developers, and implementers. This session, led by Elke Phelps, Senior Principal Product Manager, and Santiago Bastidas, Principal Product Manager, will cover best practices, tools, utilities, and tasks to minimize your maintenance downtimes during the four key maintenance phases. Topics will include: Pre-Patching: Reviewing the list of patches and analyzing their impact Patching Trials: Testing the patch prior to actual production deployment Patch Deployment: Applying patching to your system Post Patching Analysis: Validating the patch application Date:                Thursday, December 6, 2012Time:               8:00 AM - 9:00 AM Pacific Standard TimePresenters:   Elke Phelps, Senior Principal Product Manager                         Santiago Bastidas, Principal Product Manager Webcast Registration Link (Preregistration is optional but encouraged) To hear the audio feed:    Domestic Participant Dial-In Number:           877-697-8128    International Participant Dial-In Number:      706-634-9568    Additional International Dial-In Numbers Link:    Dial-In Passcode:                                              103200To see the presentation:    The Direct Access Web Conference details are:    Website URL: https://ouweb.webex.com    Meeting Number:  595757500 If you miss the webcast, or you have missed any webcast, don't worry -- we'll post links to the recording as soon as it's available from Oracle University.  You can monitor this blog for pointers to the replay. And, you can find our archive of our past webcasts and training here. If you have any questions or comments, feel free to email Bill Sawyer (Senior Manager, Applications Technology Curriculum) at BilldotSawyer-AT-Oracle-DOT-com.

    Read the article

  • SAP NetWeaver Cloud Java EE 6 Web Profile Certified!

    - by reza_rahman
    We are very pleased to welcome SAP NetWeaver Cloud to the Java EE 6 family! SAP successfully certified NetWeaver Cloud SDK-2.x.Beta against the Java EE 6 Web Profile TCK. This brings the number of Web Profile implementations to no less than seven and the total number of certified platforms on the official Java EE compatibility page to eighteen. Other Java EE 6 Web Profile platforms include the likes of GlassFish, JBoss AS, Resin and Apache TomEE. Under the hood, SAP NetWeaver Cloud uses EclipseLink, Tomcat and OpenEJB. The NetWeaver team encourages you to try it out and send them feedback. More details here.

    Read the article

  • Enable Configurator for Return Orders

    - by ChristineS-Oracle
    With release 12.2.4 for non-referenced RMAs, Order Management will allow you to configure the model from Sales Order / Quick Sales Order windows. This is only allowable when profile  OM: Enable Configuration UI for RMA is set to Yes.  All selected options must be returnable, as well as all included items. Order Management explodes included items and creates options and option classes in a way similar to outbound orders. The application creates all selected components with same line number but different option/component number.  Additionally, the application does not allow re-configuration and/or deletion of any line if any line in the same configuration is received, fulfilled, closed, cancelled, or split. For additional information refer to the Oracle Order Management Release Notes for Release 12.2.4 (Doc ID 1906521.1).

    Read the article

  • Looking for full web based booking system

    - by Sandro Dzneladze
    I've searched around here but could not find any booking system that would suit my need. I hope maybe you can help me out. I've already seen question #11379 over here. Solutions provided there are not helping and scripts not interesting. I'm looking for a booking system that support infinite number of hotels/rooms/pricingschemes. There should be unlimited number of hotel owner/users, who can access control panel and set availability for their hotel. Or create seasonal prices and adjust them. I should be able to have 15% booking fee paid through credit cart. I can work on integration with my system, it just needs to have correct APi/functionality to support this. I love wordpress so if there are some nice plugins for that, I'm open for suggestions. But this is not a must. It should be php/mysql based as I'm not good at anything else :)

    Read the article

  • Why does the first partition start at sector 34 when I choose "Guided - Use entire disk" during install?

    - by Kent
    After choosing "Guided - Use entire disk" during installation I find that the first partition starts on sector 34. Why that specific sector and not the first one? (parted) print Model: ATA WDC WD30EZRX-00M (scsi) Disk /dev/sda: 5860533168s Sector size (logical/physical): 512B/4096B Partition Table: gpt Number Start End Size File system Name Flags 1 34s 390659s 390626s fat32 boot 2 390660s 890660s 500001s ext2 3 890661s 5860533118s 5859642458s (parted) In case you prefer bytes as the unit: (parted) unit B (parted) print Model: ATA WDC WD30EZRX-00M (scsi) Disk /dev/sda: 3000592982016B Sector size (logical/physical): 512B/4096B Partition Table: gpt Number Start End Size File system Name Flags 1 17408B 200017919B 200000512B fat32 boot 2 200017920B 456018431B 256000512B ext2 3 456018432B 3000592956927B 3000136938496B

    Read the article

  • Over 2000 Windows 8 Store Apps

    - by David Paquette
    With still a month left until Windows 8 is made commercially available, I was surprised to hear yesterday that there are already 16 million devices running Windows 8 (via @TommyLee).  I was also surprised to see that in Canada, there are already over 2000 apps available in the Windows Store.  This might not sound like much, but it is double the number of apps available less than a month ago.  These look like good signs for the Windows 8 ecosystem.  I am hoping to see the number of apps continue to grow quickly between now and official launch (and beyond).

    Read the article

  • RAID1: can't replace faulty spare (marked again as 'faulty spare' within seconds)

    - by user212475
    I got a problem that I cannot solve: Our fileserver runs XUbuntu and 3 RAID1s. One has a problem since monday: it consists of sdb and sdc. sdb was marked as faulty by mdadm for unknown reasons. I used --remove to remove it from the RAID and then to add it by --add. All was fine, re-syncing started but never got above 0% and after a few seconds, sdb was again marked as 'faulty spare' (and therefore the RAID degraded, but clean). So I saved the first 512 byte of the old sdb to a file, bought a new HDD of same size (4TB), shut down the computer and replaced sdb physically, switched the computer back on and wrote the 512 byte back to the new drive to have the same partition info as the old drive (both are the same type, from same company). But the new drive shows the same behaviour as the old: I can add, re-syncing starts and after a few seconds its marked as 'faulty spare'. Here exactly what i did: mdadm --remove /dev/md/1 /dev/sdb maadm --detail /dev/md/1 gives me: /dev/md/1: Version : 1.2 Creation Time : Sat Jun 8 22:32:05 2013 Raid Level : raid1 Array Size : 3906887360 (3725.90 GiB 4000.65 GB) Used Dev Size : 3906887360 (3725.90 GiB 4000.65 GB) Raid Devices : 2 Total Devices : 1 Persistence : Superblock is persistent Update Time : Thu Nov 7 06:56:13 2013 State : clean, degraded Active Devices : 1 Working Devices : 1 Failed Devices : 0 Spare Devices : 0 Name : File-Server:1 (local to host File-Server) UUID : 44ed561f:b733e946:e69820f4:aba9b223 Events : 2424 Number Major Minor RaidDevice State 0 0 0 0 removed 1 8 32 1 active sync /dev/sdc mdadm --add /dev/md/1 /dev/sdb mdadm --detail /dev/md/1 gives me: Version : 1.2 Creation Time : Sat Jun 8 22:32:05 2013 Raid Level : raid1 Array Size : 3906887360 (3725.90 GiB 4000.65 GB) Used Dev Size : 3906887360 (3725.90 GiB 4000.65 GB) Raid Devices : 2 Total Devices : 2 Persistence : Superblock is persistent Update Time : Thu Nov 7 06:57:49 2013 State : clean, degraded, recovering Active Devices : 1 Working Devices : 1 Failed Devices : 1 Spare Devices : 0 Rebuild Status : 0% complete Name : File-Server:1 (local to host File-Server) UUID : 44ed561f:b733e946:e69820f4:aba9b223 Events : 2431 Number Major Minor RaidDevice State 2 8 16 0 faulty spare rebuilding /dev/sdb 1 8 32 1 active sync /dev/sdc and after a few seconds: /dev/md/1: Version : 1.2 Creation Time : Sat Jun 8 22:32:05 2013 Raid Level : raid1 Array Size : 3906887360 (3725.90 GiB 4000.65 GB) Used Dev Size : 3906887360 (3725.90 GiB 4000.65 GB) Raid Devices : 2 Total Devices : 2 Persistence : Superblock is persistent Update Time : Thu Nov 7 06:57:50 2013 State : clean, degraded Active Devices : 1 Working Devices : 1 Failed Devices : 1 Spare Devices : 0 Name : File-Server:1 (local to host File-Server) UUID : 44ed561f:b733e946:e69820f4:aba9b223 Events : 2436 Number Major Minor RaidDevice State 0 0 0 0 removed 1 8 32 1 active sync /dev/sdc 2 8 16 - faulty spare /dev/sdb same behaviour if I zero the superblock (mdadm --zero-superblock /dev/sdb) before adding sdb. I do all commands as root and the system holds 3 more 4TB drives, ie the mainboard can handle them. The old harddrive was checked for errors using badblocks, but all is fine. Does anybody have any idea, what the problem is?

    Read the article

  • I'm having trouble understanding these exercises wording.

    - by KasHKoW
    Exercise 1-20. Write a program detab that replaces tabs in the input with the proper number of blanks to space to the next tab stop. Assume a fixed set of tab stops, say every n columns. Should n be a variable or a symbolic parameter? Exercise 1-21. Write a program entab that replaces strings of blanks by the minimum number of tabs and blanks to achieve the same spacing. Use the same tab stops as for detab. When either a tab or a single blank would suffice to reach a tab stop, which should be given preference? could you paraphrase these for me. thanks

    Read the article

  • Sprite/Tile Sheets Vs Single Textures

    - by Reanimation
    I'm making a race circuit which is constructed using various textures. To provide some background, I'm writing it in C++ and creating quads with OpenGL to which I assign a loaded .raw texture too. Currently I use 23 500px x 500px textures of which are all loaded and freed individually. I have now combined them all into a single sprite/tile sheet making it 3000 x 2000 pixels seems the number of textures/tiles I'm using is increasing. Now I'm wondering if it's more efficient to load them individually or write extra code to extract a certain tile from the sheet? Is it better to load the sheet, then extract 23 tiles and store them from one sheet, or load the sheet each time and crop it to the correct tile? There seems to be a number of way to implement it... Thanks in advance.

    Read the article

  • How can I create multiple mini-sites with similar/duplicate content without hurting my search engine rank?

    - by ekpyrotic
    Essential background: I run a small company that lets members of the public post handwritten letters to their local politician (UK-based). Every week a number of early stage bills (called Early Day Motions) are submitted for debate in the House of Commons, and supporters of the motion will contact their local Members of Parliament, asking them to sign the motion. The crux: I want to target these EDMs with customised mini-sites, so when people search "EDM xxx", they find my customised mini-site, specifically targeting that EDM (i.e., "Send a handwritten letter to your MP asking them to sign EDM xxx"). At the moment, all these mini-sites (and my homepage) have duplicate content with only the relevant EDM name, number, and background image changed. (For example, http://mailmymp.com and http://mailmymp.com/edm/teaching-life-saving-skills-at-school-edm-550.php). The question: Firstly, will this hurt my potential search engine ranking? And, if so, what's the best way to target these political campaigns in an efficient manner without hurting my SEO prospects?

    Read the article

  • ATG Live Webcast Dec. 13th: EBS Future Directions: Deployment and System Administration

    - by Bill Sawyer
    This webcast provides an overview of the improvements to Oracle E-Business Suite deployment and system administration that are planned for the upcoming EBS 12.2 release.   It is targeted to system administrators, DBAs, developers, and implementers. This webcast, led by Max Arderius, Manager Applications Technology Group, compares existing deployment and system administration tools for EBS 12.0 and 12.1 with the upcoming functionality planned for EBS 12.2. This was a very popular session at OpenWorld 2012, and I am pleased to bring it to the ATG Live Webcast series.  This session will cover: Understanding the Oracle E-Business Suite 12.2 Architecture Installing & Upgrading EBS 12.2 Online Patching in EBS 12.2 Cloning in EBS 12.2 Date:             Thursday, December 13, 2012Time:             8:00 AM - 9:00 AM Pacific Standard TimePresenter:   Max Arderius, Manager Applications Technology Group Webcast Registration Link (Preregistration is optional but encouraged) To hear the audio feed:   Domestic Participant Dial-In Number:           877-697-8128    International Participant Dial-In Number:      706-634-9568    Additional International Dial-In Numbers Link:    Dial-In Passcode:                                              103194To see the presentation:    The Direct Access Web Conference details are:    Website URL: https://ouweb.webex.com    Meeting Number:  593672805If you miss the webcast, or you have missed any webcast, don't worry -- we'll post links to the recording as soon as it's available from Oracle University.  You can monitor this blog for pointers to the replay. And, you can find our archive of our past webcasts and training here. If you have any questions or comments, feel free to email Bill Sawyer (Senior Manager, Applications Technology Curriculum) at BilldotSawyer-AT-Oracle-DOT-com.

    Read the article

  • Implementing a Custom Coherence PartitionAssignmentStrategy

    - by jpurdy
    A recent A-Team engagement required the development of a custom PartitionAssignmentStrategy (PAS). By way of background, a PAS is an implementation of a Java interface that controls how a Coherence partitioned cache service assigns partitions (primary and backup copies) across the available set of storage-enabled members. While seemingly straightforward, this is actually a very difficult problem to solve. Traditionally, Coherence used a distributed algorithm spread across the cache servers (and as of Coherence 3.7, this is still the default implementation). With the introduction of the PAS interface, the model of operation was changed so that the logic would run solely in the cache service senior member. Obviously, this makes the development of a custom PAS vastly less complex, and in practice does not introduce a significant single point of failure/bottleneck. Note that Coherence ships with a default PAS implementation but it is not used by default. Further, custom PAS implementations are uncommon (this engagement was the first custom implementation that we know of). The particular implementation mentioned above also faced challenges related to managing multiple backup copies but that won't be discussed here. There were a few challenges that arose during design and implementation: Naive algorithms had an unreasonable upper bound of computational cost. There was significant complexity associated with configurations where the member count varied significantly between physical machines. Most of the complexity of a PAS is related to rebalancing, not initial assignment (which is usually fairly simple). A custom PAS may need to solve several problems simultaneously, such as: Ensuring that each member has a similar number of primary and backup partitions (e.g. each member has the same number of primary and backup partitions) Ensuring that each member carries similar responsibility (e.g. the most heavily loaded member has no more than one partition more than the least loaded). Ensuring that each partition is on the same member as a corresponding local resource (e.g. for applications that use partitioning across message queues, to ensure that each partition is collocated with its corresponding message queue). Ensuring that a given member holds no more than a given number of partitions (e.g. no member has more than 10 partitions) Ensuring that backups are placed far enough away from the primaries (e.g. on a different physical machine or a different blade enclosure) Achieving the above goals while ensuring that partition movement is minimized. These objectives can be even more complicated when the topology of the cluster is irregular. For example, if multiple cluster members may exist on each physical machine, then clearly the possibility exists that at certain points (e.g. following a member failure), the number of members on each machine may vary, in certain cases significantly so. Consider the case where there are three physical machines, with 3, 3 and 9 members each (respectively). This introduces complexity since the backups for the 9 members on the the largest machine must be spread across the other 6 members (to ensure placement on different physical machines), preventing an even distribution. For any given problem like this, there are usually reasonable compromises available, but the key point is that objectives may conflict under extreme (but not at all unlikely) circumstances. The most obvious general purpose partition assignment algorithm (possibly the only general purpose one) is to define a scoring function for a given mapping of partitions to members, and then apply that function to each possible permutation, selecting the most optimal permutation. This would result in N! (factorial) evaluations of the scoring function. This is clearly impractical for all but the smallest values of N (e.g. a partition count in the single digits). It's difficult to prove that more efficient general purpose algorithms don't exist, but the key take away from this is that algorithms will tend to either have exorbitant worst case performance or may fail to find optimal solutions (or both) -- it is very important to be able to show that worst case performance is acceptable. This quickly leads to the conclusion that the problem must be further constrained, perhaps by limiting functionality or by using domain-specific optimizations. Unfortunately, it can be very difficult to design these more focused algorithms. In the specific case mentioned, we constrained the solution space to very small clusters (in terms of machine count) with small partition counts and supported exactly two backup copies, and accepted the fact that partition movement could potentially be significant (preferring to solve that issue through brute force). We then used the out-of-the-box PAS implementation as a fallback, delegating to it for configurations that were not supported by our algorithm. Our experience was that the PAS interface is quite usable, but there are intrinsic challenges to designing PAS implementations that should be very carefully evaluated before committing to that approach.

    Read the article

  • Coherence Query Performance in Large Clusters

    - by jpurdy
    Large clusters (measured in terms of the number of storage-enabled members participating in the largest cache services) may introduce challenges when issuing queries. There is no particular cluster size threshold for this, rather a gradually increasing tendency for issues to arise. The most obvious challenges are that a client's perceived query latency will be determined by the slowest responder (more likely to be a factor in larger clusters) as well as the fact that adding additional cache servers will not increase query throughput if the query processing is not compute-bound (which would generally be the case for most indexed queries). If the data set can take advantage of the partition affinity features of Coherence, then the application can use a PartitionedFilter to target a query to a single server (using partition affinity to ensure that all data is in a single partition). If this can not be done, then avoiding an excessive number of cache server JVMs will help, as will ensuring that each cache server has sufficient CPU resources available and is also properly configured to minimize GC pauses (the most common cause of a slow-responding cache server).

    Read the article

< Previous Page | 260 261 262 263 264 265 266 267 268 269 270 271  | Next Page >