Search Results

Search found 5602 results on 225 pages for 'tuning red gate'.

Page 121/225 | < Previous Page | 117 118 119 120 121 122 123 124 125 126 127 128  | Next Page >

• compile ntp without ssl

  - by Zulakis

  I need to deploy ntp to a very space-critical pxe-imaging-system. (Yes, each KB matters.) Footprint needs to be as small as possible, so I want to compile ntp without linking openssl. According to the manual this is should be possible: If available, the OpenSSL library from http://www.openssl.org is used to support public key cryptography. The library must be built and installed prior to building NTP. The procedures for doing that are included in the OpenSSL documentation. The library is found during the normal NTP configure phase and the interface routines compiled automatically. Only the libcrypto.a library file and openssl header files are needed. If the library is not available or disabled, this step is not required. I already tried out ./configure --without-openssl however, this didn't help. This is my ldd output: ldd ntpd/ntpd linux-gate.so.1 => (0xb7706000) libm.so.6 => /lib/i686/cmov/libm.so.6 (0xb76d5000) libcrypto.so.0.9.8 => /usr/lib/i686/cmov/libcrypto.so.0.9.8 (0xb7582000) librt.so.1 => /lib/i686/cmov/librt.so.1 (0xb7578000) libc.so.6 => /lib/i686/cmov/libc.so.6 (0xb741d000) /lib/ld-linux.so.2 (0xb7707000) libdl.so.2 => /lib/i686/cmov/libdl.so.2 (0xb7419000) libz.so.1 => /usr/lib/libz.so.1 (0xb7404000) libpthread.so.0 => /lib/i686/cmov/libpthread.so.0 (0xb73eb000) The system I am compiling on is 32-bit debian lenny using openssl 0.9.8g-15+lenny16. What is the correct configure option to compile ntp without openssl?

  Read the article

• overriding default scheduler for blkio requests in cgroups

  - by Aamir Mushtaq

  I am trying to optimize a set of servers that have to reside on single machine. i.e. i can have multiple application server, a DB server and of course a samba server as well in same instance. Now I was looking into several optimizing options available to me. In my quest, i did my tuning of the network stack. coming to the CPU, MEMORY and the BLKIO tweaks, i am using CGROUPS. The problem i am facing is that for enhanced performance in the nature of the applications that i am running, the CFQ Scheduler that is implemented for the BLKIO subsystem is not optimal. I was looking more for a Deadline Scheduler because that will serve my purpose well. My question is whether it is possible for us to change the scheduler in the kernel compilation itself for the BLKIO to Deadline and it will reflect in my usage of [CGROUP hierarchies][3]? Since when running the service cgconf, a new fs is mounted and i dont want it to revert to CFQ scheduler. I also welcome any suggestions that will enable me to have more control over my resources.

  Read the article

• SQL Server performance on VSphere 4.0

  - by Charles

  We are having a performance issue that we cannot explain with our VMWare environment and I am hoping someone here may be able to help. We have a web application that uses a databases backend. We have an SQL 2005 Cluster setup on Windows 2003 R2 between a physical node and a virtual node. Both physical servers are identical 2950's with 2x Xeaon x5460 Quad Core CPUs and 64GB of memory, 16GB allocated to the OS. We are utilizing an iSCSI San for all cluster disks. The problem is this, when utilizing the application under a repeated stress testing that adds CPUs to the cluster nodes, the Physical node scales from 1 pCPU to 8 pCPUs, meaning we see continued performance increases. When testing the node running Vsphere, we have the expected 12% performance hit for being virtual but we still scale from 1 vCPU to 4 vCPUs like the physical but beyond this performance drops off, by the time we get to 8 vCPUs we are seeing performance numbers worse than at 4 vCPUs. Again, both nodes are configured identically in terms of hardware, Guest OS, SQL Configurations etc and there is no traffic other than the testing on the system. There are no other VMs on the virtual server so there should be no competition for resources. We have contacted VMWare for help but they have not really been any suggesting things like setting SQL Processor Affinity which, while being helpful would have the same net effect on each box and should not change our results in the least. We have looked at all of VMWare's SQL Tuning guides with regards to VSphere with no benefit, please help!

  Read the article

• Slow Local Network, Windows 7, Snow Leopard, WiFi/Wired

  - by WerkkreW

  I am experiencing really poor local network performance in my home. I was recently using a Linksys WRT54G Router with DD-WRT on it, and a couple comparable Linksys-G PCI cards for connectivity but decided to upgrade hoping it would help with my performance issues. The computers in my house are connected as follows: Comcast Business Class Commercial 25mbps/10mbps (Verified) D-Link DGL-4500 Wireless N Router Windows 7x64 - D-Link DWA-552 Wireless-N Windows 7x64 - D-Link DWA-552 Wireless-N Mac Mini 10.6.2 - AirPort Extreme N Playstation 3, Hard Wired Xbox 360, Hard Wired Essentially the problem is very specific. Web browsing and uploading/downloading files from the internet is fine, more than fine. But if I want to say, Stream a video from one of my Windows 7 computers to my PS3, or copy a large video file between either of the PC's or the Mac, I get a consistent 500-900Kbps throughput at the high end. If I open my network browser, or try to browse my homegroup the response time is horrible. Both of my Windows computers are showing Strong wireless signals with a connection speed of 300Mbps. I know I can never expect to achieve anything near those speeds, but 500Kbps? Here is what I have tried so far: Enabled Single mode N-only and N/G Only on router WPA2 with AES Encrpytion Disabled "Remote Differential Compression" in Windows 7 Disabled TCP "Auto-Tuning" Used other software for file copies such as "Teracopy" I am at the end of my rope. Unfortunately I live in a 75 year old home with plaster walls, so hard-wiring my entire house isn't really an option I can handle right now. Any ideas to help me get decent speed when transferring files across my network would be greatly appreciated.

  Read the article

• Is there any way to limit the turbo boost speed / intensity on i7 lap?

  - by Anonymous

  I've just got a used i7 laptop, one of these overheating pavilions from HP with quad cores. And I really want to find a compromise between the temp and performance. If I use linpack, or some other heavy benchmark, the temp easily gets to 95+, and having a TJ of 100 Degrees, for a 2630QM model, it really gets me throttling, that no cooling pad or even an industrial fan could solve. I figured later that it is due to turbo boost, and if I set my power settings to use 99% of the CPU instead of 100%, and it seems to disable the turbo boost, so the temp gets better. But then again it loses quite a bit of performance. The regular clock is 2GHz, and in turbo boost it gets to 2.6Ghz, but I just wonder if I could limit it to around 2.3Ghz, that would be a real nice thing. Also there is another question I've hard time getting answer to. It seems to me that clocks are very quickly boosting up to max even when not needed, eg, it's ok if the CPU has 0% load, the clocks get to their 800MHz, but even if it gets to about 5% it quickly jumps to a max and even popping up turbo, which seems very strange to me. So I wonder if there is any way to adjust the sensitivity of the Speed Step feature. I believe it would be more logical to demand increased clock if it hits let's say 50% load. I do understand that most of these features are probably hardwired somewhere in the CPU itself or the MB, which has no tuning options just like on many laptops. But I would appreciate if you could recommend some thing, or some software. Thanks

  Read the article

• Caching all files in varnish

  - by csgwro

  I want my varnish servers to cache all files. At backend there is lighttpd hosting only static files, and there is an md5 in the url in case of file change, ex. /gfx/Bird.b6e0bc2d6cbb7dfe1a52bc45dd2b05c4.swf). However my hit ratio is very poorly (about 0.18) My config: sub vcl_recv { set req.backend=default; ### passing health to backend if (req.url ~ "^/health.html$") { return (pass); } remove req.http.If-None-Match; remove req.http.cookie; remove req.http.authenticate; if (req.request == "GET") { return (lookup); } } sub vcl_fetch { ### do not cache wrong codes if (beresp.status == 404 || beresp.status >= 500) { set beresp.ttl = 0s; } remove beresp.http.Etag; remove beresp.http.Last-Modified; } sub vcl_deliver { set resp.http.expires = "Thu, 31 Dec 2037 23:55:55 GMT"; } I have made an performance tuning: DAEMON_OPTS="${DAEMON_OPTS} -p thread_pool_min=200 -p thread_pool_max=4000 -p thread_pool_add_delay=2 -p session_linger=100" The main url which is missed is... /health.html. Is that forward to backend correctly configured? Disabling health checking hit ratio increases to 0.45. Now mostly "/crossdomain.xml" is missed (from many domains, as it is wildcard). How can I avoid that? Should I carry on other headers like User-Agent or Accept-Encoding? I thing that default hashing mechanism is using url + host/IP. Compression is used at the backend. What else can improve performance?

  Read the article

• Telugu (unicode) font rendering in emacs

  - by Prakash K

  [I asked the following question in stackoverflow, and I have been redirected here. I hope I can get some answers here. My question at stackoverflow had two small images showing the example rendering of text. As a new user at superuser, I am not being allowed to include them here, nor I am allowed to post more than one hyperlink. And, I don't have enough reputation on SO to migrate that question. Please look at the stackoverflow question for the images. Sorry about the inconvenience.] I sometimes edit text in telugu language. However, when I open the file (UTF-8 encoded) in GNU emacs (version 23.1.50.1 on Ubuntu Jaunty) the text rendering is incorrect. The same text file opened in gedit is rendered correctly. Here's a snippet: ????????? ???? ???? ???????? rendred in gedit: Please see the SO question for the image showing telugu text rendering in gedit And, the emacs rendering of the same text: Please see the SO question for the image showing telugu text rendering in emacs Wherever glyphs need to be composited (not sure if it's the right word), emacs (or whatever library it uses) is not doing it right. Is there anyway to fix this? Perhaps tuning some setting in my configuration? Any ideas, please?

  Read the article

• SSD as primary or secondary drive on a small Linux server?

  - by Alex Martelli

  I'm pensioning off my 10-years-old home server and replacing it with an Ubuntu 10.04 box. The two storage devices are a Western Digital Caviar Green 2.0TB HD and an Intel X25-M 34nm Gen 2 80GB SATA II 2.5inch SSD (the box has 8GB RAM and an i5 750, if it matters). I don't care much about boot times (since I don't plan to reboot all that often;-); the main frequent, performance-demanding task will be (re)building large open source C or C++ software packages from sources (as an open source contributor, I do that often). So, I thought I'd keep the SSD as the secondary drive and the HD as the primary one, using the SSD mostly for the files that can otherwise demand a lot of seeking (esp. in a parallel make). However, the friendly vendor (perhaps more experienced in Windows systems than in Linux ones) thinks the "normal" way to configure the machine would be with the SSD as the primary drive. I'm pretty rusty on configuring and tuning systems, so, I thought I'd better double check on SuperUser... thanks in advance for advice about this choice!

  Read the article

• Unable to set initcwnd on a Hetzner server

  - by Sergi

  We just ordered a bunch of Hetzner EX40SSD servers with the minimal Debian install image that they provide and everything is just fine except that looking at tcpdumps for fine tuning the network from various locations the initcwnd param seems to be stuck at 6 no matter how we change it. By default Debian 3.2 kernels should have that setting to 10 so it's pretty strange. Is it possible that the NIC driver or a custom setting in the Hetzner Debian image is limiting this param? Even if we set it to 4, like the old kernel default, it doesn't work. Any ideas would be much appreciated! Does anyone know if the NIC drivers provided by default by Debian have some kind on limitation. In a long thread in http://www.webhostingtalk.com/showthread.php?t=1200617&highlight=hetzner they talk about a page http://wiki.hetzner.de/index.php/Installation_des_r8168-Treibers/en where Hetzner states that the included Realtek r8168 driver is not working properly, but nowhere do they say that the initcwnd could be affected. Tomorrow i will try to install a CentOs image and see if Debian is the problem...Last resort would be to install a custom debian image, but that is a pain in the ass! Thanks!

  Read the article

• Connecting to Server 2008 shares fails

  - by Chris J

  I'm having problems getting a reliable share working on an x64 Server 2008 R1 SP1 server. All works well after a reboot, but after some time (within a day) the shares become unavailable to XP and Server 2003 servers. Interestingly, they remain available to other Server 2008 servers. On trying to access \\server\share, Server 2003 returns immediately and simply gives me the message "The specified network name is no longer available", XP takes a minute or two to timeout before giving the same message. There doesn't seem to be anything in the event logs indicating a problem. Doing some googling over the last day or two I've seen the following blamed: Bad network drivers ... I've updated to the latest drivers with no result Symantec anti-virus ... we're not using it (currently no AV on the server) Receive window auto-tuning ... I've disabled with netsh int tcp set global autotuninglevel=disabled and netsh int tcp set global rss=disabled None of these have had an effect. Windows Firewall is currently disabled. As other Server 2008 boxes (both x32 and x64) can connect, I can only assume that there's some new security configuration that's not quite right - or there's an AD issue that I need to trace, but don't know where to start. Even if anyone doesn't know how to resolve, if someone knows what I need to look for with Wireshark this would be a help.

  Read the article

• Bypass network stack. Which options do we have? Pros and cons of each option [on hold]

  - by javapowered

  I'm writing trading application. I want to bypass network stack in Linux but I don't know how this can be done. I'm looking for complete list of options with pros and cons of each of them. The only option I know - is to buy solarflare network card which supports OpenOnLoad. What other options should I consider and what is pros and cons of each of them? Well the question is pretty simple - what is the best way to bypass network stack? upd: OpenOnload It achieves performance improvements in part by performing network processing at user-level, bypassing the OS kernel entirely on the data path. Intel DDIO to allow Intel® Ethernet Controllers and adapters to talk directly with the processor cache of the Intel® Xeon® processor E5. What's key difference between these techologies? Do they do roughly the same things? I much better like Intel DDIO because it's much easy to use, but OpenOnload required a lot of installation and tuning. If good OpenOnload application is much faster than good Intel DDIO application?

  Read the article

• Redis connection issue

  - by mre

  We are currently experiencing a lot of Redis errors with the message Unable to connect: read error on connection, trying next server We run Redis on FreeBSD using PHP Redis and we have a hard time reproducing the error on Ubuntu so this might be a hint. There's a long-running issue on that topic on github. Basically we get a socket from the operating system with a call to connect(host, port, timeout) in phpredis, but when we do a select(db_index) afterwards, we get an exception. Could there be an issue with persistance? I assume that connect does nothing in the background and select tries to access the connection, which is actually closed. We don't run into a timeout. We tried tuning TIME_WAIT without success. Any other ideas on where the problem might come from? What is the best way to track the issue down? dtrace maybe? Update We are currently looking into our BGSAVE settings. Interestingly it takes half a second and more to create a fork for the process which regularly writes the data to disk (persistence) and maybe redis can't respond to connect() requests during that timespan.

  Read the article

• Ubuntu Server Wireless connection issue - replaced router but kept ESSID

  - by Stevo

  I have a ubuntu server 12.04 which was connected to my wireless network with no problem I replaced the wireless router but kept the ESSID and password the same. All other devices on network have connected correctly. However the Ubuntu Server will not route correctly. It will connect to the wifi router, and get a dhcp served IP address, however it will not route anything. I cannot ping the router from the server. the contents of /etc/resolve.conf are updated with the information from the router, (the host name has been served) I know there is nothing wrong with the router or the server, or the wireless card etc. I'm assuming there's some cached setting that associates the old router with the ESSID and causing the issue. I've got a lot of other devices connected to the router, so don't want to change the name of the ESSID. How do I fix this? EDIT: outputs (abbreviated as I've got no cut and paste) netstat -rn: Kernel IP Routing table Dest Gate Gen Flags MSS Window irtt Iface 0.0.0.0 192.168.0.1 0.0.0.0. UG 0 0 0 wlan0 192.168.0.0 0.0.0.0. 255.255.255.0 U 0 0 0 wlan0

  Read the article

• What does SQL Server's BACKUPIO wait type mean?

  - by solublefish

  I'm using Sql Server 2008 ("R1"), with some maintenance plans that back up my databases to a network share. Some of my backup jobs show long waits of type "BACKUPIO". Of course it seems like this is an I/O subsystem limitation, but I'm skeptical. Perfmon stats for I/O on the production (source) server are well within normal trends for that server. The destination server shows a sustained 7MB/s write rate, which seems incredibly low, even for a slow disk. The network link is gigabit ethernet and nowhere near saturated. The few docs I've turned up about BACKUPIO indicate that it's not specifically a wait on I/O, surprisingly enough. This MSFT doc says it's abnormal unless you're using a tape drive, which I'm not. But it doesn't say (or I don't understand) exactly what resource is missing. http://www.docstoc.com/docs/24580659/Performance-Tuning-in-SQL-Server-2005 And this piece says it's not related to I/O performance at all. http://www.informit.com/articles/article.aspx?p=686168&seqNum=5 "Note that BACKUPIO and IO_AUDIT_MUTEX are not related to IO performance." Anyway, does anyone know what BACKUPIO actually means and/or what I can do to diagnose or eliminate it?

  Read the article

• MySQL Memory Limit Windows Server 2003

  - by Matt

  I am running MySQL 5.0.51a on Windows Server 2003 Standard Edition on an HP DL580 G4 with 3GB installed. One of my database tables has grown to 5.3 GB with an index file of 2.5 GB, which I believe is causing MySQL to be slow due to having to constantly load and unload the index file when updates are made to the table. The server itself seems to be performing OK because MySQL is only using about 500MB of memory (there are other apps running on the system, but MySQL uses the most memory). The table is fairly active with new records getting adding all during day but no deletes, ever. The MySQL server has up to 600 connections allowed, but only small number (10 or 20) would actually be writing to this table. I increased the memory limits in MySQL but since the max connections is so high I don't think I can give each connection 1GB without risking a problem. Is there some tuning that would let just certain connections get a lot of memory? So I have started to look for alternatives to avert the crisis I know is coming soon. Some of the options I have: Upgrade to Server 2003 Enterprise to install 64GB of memory. Question: would 32 bit MySQL be able to access more than 2GB? Would that be 2GB per thread? That would still be smaller than the index table size so it might not solve the problem completely, but it would be better than now. Upgrade to Server 200x 64 bit and MySQL 64 bit. Switch to a *nix 64 bit server. If anybody has suggestions for things to do in the meantime, opinions on which way to go, or other things that I have overlooked I would appreciate the help. Thanks

  Read the article

• Why is MySQL unable to open hosts.allow/hosts.deny?

  - by HonoredMule

  I have a storage server running Nexenta (OpenSolaris kernel, Ubuntu userspace) with MySQL on top of a ZFS storage array, using innodb_file_per_table and ulimit -n set to 8K. mysqltuner.pl confirms the file limit and claims there are 169 files. The following command: pfiles `fuser -c / 2>/dev/null indicates one mysqld process having 485 file/device descriptors (and they're almost all for files) so I don't know how reliable the tuning script is, but it is still way less than 8K and this list also finds no other process which is close to it's limit. The global total number of descriptors in use is around 1K. So what can cause mysqld to be constantly streaming the following errors? [date] [host] mysqld[pid]: warning: cannot open /etc/hosts.allow: Too many open files [date] [host] mysqld[pid]: warning: cannot open /etc/hosts.deny: Too many open files Everything appears to actually be operating fine, but the issue is constantly flooding the admin console and starts right away on a fresh boot (not only reproducible, but always from mysqld and always the hosts files, whose permissions are the default -rw-r--r-- 1 root root). I could, of course, suppress it from the admin console but I'd rather get to the bottom of it and still allow mysqld warnings/errors to reach the admin console. EDIT: not only is the actual file descriptor well within sane limits, the issue also persists (with immediate appearance) even with the file limit raised to 65535 and always only on hosts.allow/deny.

  Read the article

• Despeckle line art

  - by Dour High Arch

  We have a number of line-art charts unfortunately saved as JPEGs. They are now riddled with distracting compression artifacts or "speckles". Is there any way of removing these? I do not have the original files and it will be very difficult to recreate them. I am running Windows 7 and tried Paint.Net; none of the filters help. Posterize washed out all the colors and leaves the speckles. Blur makes text unreadable. Noise Reduction wrecks antialiasing of curved lines, and perversely enhances the speckles, making them look like checkerboards. Yes, I have Googled for software to do this; there are many programs that advertise despeckling but, after my experience with Paint.Net, do not want to experiment with applications that show no before and after images. The only example I have seen that does what I want is from a Photoshop tutorial. I have dozens of files and the tutorial requires considerable manual fine-tuning. I would prefer to automate or batch-process this task. Commercial apps are fine, but I do not want to spend over $600 and learning a complex program for a single task.

  Read the article

• My 3D Games crashing in these scenarios

  - by desaivv

  I have a situation here which I am unable to solve. I bought a PC last year March, here are my specs: Intel Core i3 550 @ 3GHz 4 GB RAM @400 MHz XFX GeForce 9500 GT graphics card 1GB @550MHz 500 GB HDD Lately as soon as I load my save game of Skyrim, it crashes. I have been playing Skyrim since I joined Gaming.SE site. Crashes as in the entire scene gets red lines. I can not ALT + TAB back or even CTRL + ALT + DEL either. My only recourse is a hard reset via the power button. Can not take a screen shot either. I have the latest Forceware 296.10 drivers also. This has been happening since the last 2 weeks. I always use Driver Sweeper to clean my old drivers, since that is what XFXForce recommends before installing new drivers. I installed MSI Afterburner lately to see my GPU temperature. My GPU is default, never over-clocked it. In MSI's Afterburner, I can not adjust fan speed. It is greyed out. Also in settings there is no fan tab. With normal Internet browsing it stays at 51 C. Ran Memtest86 over night with level 11. Took about 13 hours, but no errors in my RAM. I even re-installed my OS, with just the 296 drivers. The fan for the GPU does come on. I can play Diablo 2. I can not get past Warcraft 3's menu selection. There WAS some dust in my machine, but I always try and keep everything clean, since in my home town dust is an issue. Always keep cool my entire PC cabinet. My friend came with his functioning graphics card, we bought our PCs at the same time with exact same specifications. His card did not work either. Same problem with the scene freezing with red lines. I did do my research before posting here. That is how I was able to learn about MSI Afterburner, Driver Sweeper, SpeedFan etc. I followed posts on Tom's Hardware too regarding people that had similar problems. One person suggested and was followed by worked as well the suggestion to "Bake the card in an oven". Since I bought it, played Diablo 2 for months, Starcraft 2 campaign for months and Skyrim recently for months. Bought ME3 also. I am at my wits end. I do not know what else to do. I can go out and buy a card, but my friend's card did not work either. I can use the machine for Eclipse or VS2010 development just fine. Just not with 3D gaming. I originally posted this question on Gaming.SE But I was directed here. I have browsed the SU database for my problem and found this, this, and this. But none of these cover my question. My machine is only one year old. Can some experienced superuser(s) shed some light on this scenario? Is it a 3D graphics card problem? Will a brand new card work? What else can I try to pin point the problem? Can it be the Motherboard? Thanks.

  Read the article

• Windows Server 2008 R2 grinds to a screeching halt during file copy operations

  - by skolima

  When my Windows Server 2008 R2 machine is performing any large disk operations (copying 10GB files from one drive to another, copying similar file over network, merging HyperV snapshots, compressing large files), performance of the whole machine slows down terribly, everything becomes unresponsive. This is noticeable in any situation when the disk access is large enough not to fit in the cache. Are there any settings available for tuning this behaviour? I can accept slower file transfer if this would give me more responsiveness. System details: Dell Optiflex 960, Core 2 Quad Q9650, 8GB RAM, 2 SATA drives - 320GB (ST3320418AS) and 1TB (ST31000528AS), NCQ active on both, Intel 82564LM-3 Gigabit Ethernet, ATI HD 3450 graphics, Intel ICH10 bridge. We have multiple machines like this, every one is exhibiting the same behaviour. I though this was overkill for a workstation, apparently I was mistaken. Update: I guess I shouldn't have mentioned the HyperV at all. The above configuration is a standard workstation setup at the company I work for, this is not a server of any kind. I have at most 3 virtual machines working, and usually I'm the only person accessing them. Never the less, the slowdown occurs even when no VMs are running. On a Linux machine I'd simply ionice the copy process and I could forget about it, is there any way to manage IO priorities on Windows?

  Read the article

• ZFS & Deduplicating FLAC Data

  - by jasongullickson

  I'm experimenting with using ZFS to deduplicate a large library of FLAC files. The purpose of this is twofold: Reduce storage utilization Reduce bandwidth needed to sync the library with cloud storage Many of these files are of the same music tracks but from different physical media. This means that for the most part they are the same and usually close to the same size, which makes me think that they should benefit from block-level deduplication. However in my testing I'm not seeing good results. When I create a pool and add three of these tracks (identical songs from different source media) zpool list reports 1.00 dedupe. If I copy all of the files (make exact duplicates of the three) dedupe climbs, so I know that it is enabled and functioning, but it's not finding any duplication in the original collection of files. My first thought was that perhaps some of the variable header data (metadata tags, etc.) might be mis-aligning the bulk of the data in these files (the audio frames) but even making the header data consistent across the three files doesn't seem to have any impact on deduplication. I'm considering taking alternate routes (testing other dedupe filesystems as well as some custom code) but since we're already using ZFS and I like the ZFS replication options, I'd prefer to use ZFS dedupe for this project; but perhaps it's simply not capable of working well with this sort of data. Any feedback regarding tuning that might improve dedupe performance for this sort of dataset, or confirmation that ZFS dedupe is not the right tool for this job are appreciated.

  Read the article

• Output current with Teensy++ 2.0 (arduino-based hardware)

  - by omtinez

  I am working on a project with a Teensy++ 2.0 for testing, eventually the goal is to use a Teensy 2.0 (info on both available here) and mount it onto a robot R/C car along with a Raspberry Pi. I have been able to use and test one of the very cheap distance sensors that use ultrasound, which requires very little current. I was trying to power a motor, I don't know exactly what kind of motor but I assume a very low-power one which is what comes with the R/C car cheapo, but nothing is happening. When I plug the motor to GROUND and +5V it runs fine, but when I use GROUND and one of the GPIO pins then nothing happens with the motor. The same GPIO pins were tested to successfully power and run the ultrasound sensor, so the board is fine. My suspicion is that the GPIO pins don't output enough current to power the motor, but my knowledge of electronics is rather scarce (I am a computer scientist, not an electrical engineer). So please forgive me if I am asking something obvious or plain stupid, but does the board not have enough power to power the motor? If so, I could try to use a second power supply that would go straight into the motor and use the GPIO as a gate to turn that power on and off; would such thing work? Is there a better design that could be used?

  Read the article

• Problems with real-valued input deep belief networks (of RBMs)

  - by Junier

  I am trying to recreate the results reported in Reducing the dimensionality of data with neural networks of autoencoding the olivetti face dataset with an adapted version of the MNIST digits matlab code, but am having some difficulty. It seems that no matter how much tweaking I do on the number of epochs, rates, or momentum the stacked RBMs are entering the fine-tuning stage with a large amount of error and consequently fail to improve much at the fine-tuning stage. I am also experiencing a similar problem on another real-valued dataset. For the first layer I am using a RBM with a smaller learning rate (as described in the paper) and with negdata = poshidstates*vishid' + repmat(visbiases,numcases,1); I'm fairly confident I am following the instructions found in the supporting material but I cannot achieve the correct errors. Is there something I am missing? See the code I'm using for real-valued visible unit RBMs below, and for the whole deep training. The rest of the code can be found here. rbmvislinear.m: epsilonw = 0.001; % Learning rate for weights epsilonvb = 0.001; % Learning rate for biases of visible units epsilonhb = 0.001; % Learning rate for biases of hidden units weightcost = 0.0002; initialmomentum = 0.5; finalmomentum = 0.9; [numcases numdims numbatches]=size(batchdata); if restart ==1, restart=0; epoch=1; % Initializing symmetric weights and biases. vishid = 0.1*randn(numdims, numhid); hidbiases = zeros(1,numhid); visbiases = zeros(1,numdims); poshidprobs = zeros(numcases,numhid); neghidprobs = zeros(numcases,numhid); posprods = zeros(numdims,numhid); negprods = zeros(numdims,numhid); vishidinc = zeros(numdims,numhid); hidbiasinc = zeros(1,numhid); visbiasinc = zeros(1,numdims); sigmainc = zeros(1,numhid); batchposhidprobs=zeros(numcases,numhid,numbatches); end for epoch = epoch:maxepoch, fprintf(1,'epoch %d\r',epoch); errsum=0; for batch = 1:numbatches, if (mod(batch,100)==0) fprintf(1,' %d ',batch); end %%%%%%%%% START POSITIVE PHASE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% data = batchdata(:,:,batch); poshidprobs = 1./(1 + exp(-data*vishid - repmat(hidbiases,numcases,1))); batchposhidprobs(:,:,batch)=poshidprobs; posprods = data' * poshidprobs; poshidact = sum(poshidprobs); posvisact = sum(data); %%%%%%%%% END OF POSITIVE PHASE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% poshidstates = poshidprobs > rand(numcases,numhid); %%%%%%%%% START NEGATIVE PHASE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% negdata = poshidstates*vishid' + repmat(visbiases,numcases,1);% + randn(numcases,numdims) if not using mean neghidprobs = 1./(1 + exp(-negdata*vishid - repmat(hidbiases,numcases,1))); negprods = negdata'*neghidprobs; neghidact = sum(neghidprobs); negvisact = sum(negdata); %%%%%%%%% END OF NEGATIVE PHASE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% err= sum(sum( (data-negdata).^2 )); errsum = err + errsum; if epoch>5, momentum=finalmomentum; else momentum=initialmomentum; end; %%%%%%%%% UPDATE WEIGHTS AND BIASES %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% vishidinc = momentum*vishidinc + ... epsilonw*( (posprods-negprods)/numcases - weightcost*vishid); visbiasinc = momentum*visbiasinc + (epsilonvb/numcases)*(posvisact-negvisact); hidbiasinc = momentum*hidbiasinc + (epsilonhb/numcases)*(poshidact-neghidact); vishid = vishid + vishidinc; visbiases = visbiases + visbiasinc; hidbiases = hidbiases + hidbiasinc; %%%%%%%%%%%%%%%% END OF UPDATES %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% end fprintf(1, '\nepoch %4i error %f \n', epoch, errsum); end dofacedeepauto.m: clear all close all maxepoch=200; %In the Science paper we use maxepoch=50, but it works just fine. numhid=2000; numpen=1000; numpen2=500; numopen=30; fprintf(1,'Pretraining a deep autoencoder. \n'); fprintf(1,'The Science paper used 50 epochs. This uses %3i \n', maxepoch); load fdata %makeFaceData; [numcases numdims numbatches]=size(batchdata); fprintf(1,'Pretraining Layer 1 with RBM: %d-%d \n',numdims,numhid); restart=1; rbmvislinear; hidrecbiases=hidbiases; save mnistvh vishid hidrecbiases visbiases; maxepoch=50; fprintf(1,'\nPretraining Layer 2 with RBM: %d-%d \n',numhid,numpen); batchdata=batchposhidprobs; numhid=numpen; restart=1; rbm; hidpen=vishid; penrecbiases=hidbiases; hidgenbiases=visbiases; save mnisthp hidpen penrecbiases hidgenbiases; fprintf(1,'\nPretraining Layer 3 with RBM: %d-%d \n',numpen,numpen2); batchdata=batchposhidprobs; numhid=numpen2; restart=1; rbm; hidpen2=vishid; penrecbiases2=hidbiases; hidgenbiases2=visbiases; save mnisthp2 hidpen2 penrecbiases2 hidgenbiases2; fprintf(1,'\nPretraining Layer 4 with RBM: %d-%d \n',numpen2,numopen); batchdata=batchposhidprobs; numhid=numopen; restart=1; rbmhidlinear; hidtop=vishid; toprecbiases=hidbiases; topgenbiases=visbiases; save mnistpo hidtop toprecbiases topgenbiases; backpropface; Thanks for your time

  Read the article

• Problems with real-valued deep belief networks (of RBMs)

  - by Junier

  I am trying to recreate the results reported in Reducing the dimensionality of data with neural networks of autoencoding the olivetti face dataset with an adapted version of the MNIST digits matlab code, but am having some difficulty. It seems that no matter how much tweaking I do on the number of epochs, rates, or momentum the stacked RBMs are entering the fine-tuning stage with a large amount of error and consequently fail to improve much at the fine-tuning stage. I am also experiencing a similar problem on another real-valued dataset. For the first layer I am using a RBM with a smaller learning rate (as described in the paper) and with negdata = poshidstates*vishid' + repmat(visbiases,numcases,1); I'm fairly confident I am following the instructions found in the supporting material but I cannot achieve the correct errors. Is there something I am missing? See the code I'm using for real-valued visible unit RBMs below, and for the whole deep training. The rest of the code can be found here. rbmvislinear.m: epsilonw = 0.001; % Learning rate for weights epsilonvb = 0.001; % Learning rate for biases of visible units epsilonhb = 0.001; % Learning rate for biases of hidden units weightcost = 0.0002; initialmomentum = 0.5; finalmomentum = 0.9; [numcases numdims numbatches]=size(batchdata); if restart ==1, restart=0; epoch=1; % Initializing symmetric weights and biases. vishid = 0.1*randn(numdims, numhid); hidbiases = zeros(1,numhid); visbiases = zeros(1,numdims); poshidprobs = zeros(numcases,numhid); neghidprobs = zeros(numcases,numhid); posprods = zeros(numdims,numhid); negprods = zeros(numdims,numhid); vishidinc = zeros(numdims,numhid); hidbiasinc = zeros(1,numhid); visbiasinc = zeros(1,numdims); sigmainc = zeros(1,numhid); batchposhidprobs=zeros(numcases,numhid,numbatches); end for epoch = epoch:maxepoch, fprintf(1,'epoch %d\r',epoch); errsum=0; for batch = 1:numbatches, if (mod(batch,100)==0) fprintf(1,' %d ',batch); end %%%%%%%%% START POSITIVE PHASE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% data = batchdata(:,:,batch); poshidprobs = 1./(1 + exp(-data*vishid - repmat(hidbiases,numcases,1))); batchposhidprobs(:,:,batch)=poshidprobs; posprods = data' * poshidprobs; poshidact = sum(poshidprobs); posvisact = sum(data); %%%%%%%%% END OF POSITIVE PHASE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% poshidstates = poshidprobs > rand(numcases,numhid); %%%%%%%%% START NEGATIVE PHASE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% negdata = poshidstates*vishid' + repmat(visbiases,numcases,1);% + randn(numcases,numdims) if not using mean neghidprobs = 1./(1 + exp(-negdata*vishid - repmat(hidbiases,numcases,1))); negprods = negdata'*neghidprobs; neghidact = sum(neghidprobs); negvisact = sum(negdata); %%%%%%%%% END OF NEGATIVE PHASE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% err= sum(sum( (data-negdata).^2 )); errsum = err + errsum; if epoch>5, momentum=finalmomentum; else momentum=initialmomentum; end; %%%%%%%%% UPDATE WEIGHTS AND BIASES %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% vishidinc = momentum*vishidinc + ... epsilonw*( (posprods-negprods)/numcases - weightcost*vishid); visbiasinc = momentum*visbiasinc + (epsilonvb/numcases)*(posvisact-negvisact); hidbiasinc = momentum*hidbiasinc + (epsilonhb/numcases)*(poshidact-neghidact); vishid = vishid + vishidinc; visbiases = visbiases + visbiasinc; hidbiases = hidbiases + hidbiasinc; %%%%%%%%%%%%%%%% END OF UPDATES %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% end fprintf(1, '\nepoch %4i error %f \n', epoch, errsum); end dofacedeepauto.m: clear all close all maxepoch=200; %In the Science paper we use maxepoch=50, but it works just fine. numhid=2000; numpen=1000; numpen2=500; numopen=30; fprintf(1,'Pretraining a deep autoencoder. \n'); fprintf(1,'The Science paper used 50 epochs. This uses %3i \n', maxepoch); load fdata %makeFaceData; [numcases numdims numbatches]=size(batchdata); fprintf(1,'Pretraining Layer 1 with RBM: %d-%d \n',numdims,numhid); restart=1; rbmvislinear; hidrecbiases=hidbiases; save mnistvh vishid hidrecbiases visbiases; maxepoch=50; fprintf(1,'\nPretraining Layer 2 with RBM: %d-%d \n',numhid,numpen); batchdata=batchposhidprobs; numhid=numpen; restart=1; rbm; hidpen=vishid; penrecbiases=hidbiases; hidgenbiases=visbiases; save mnisthp hidpen penrecbiases hidgenbiases; fprintf(1,'\nPretraining Layer 3 with RBM: %d-%d \n',numpen,numpen2); batchdata=batchposhidprobs; numhid=numpen2; restart=1; rbm; hidpen2=vishid; penrecbiases2=hidbiases; hidgenbiases2=visbiases; save mnisthp2 hidpen2 penrecbiases2 hidgenbiases2; fprintf(1,'\nPretraining Layer 4 with RBM: %d-%d \n',numpen2,numopen); batchdata=batchposhidprobs; numhid=numopen; restart=1; rbmhidlinear; hidtop=vishid; toprecbiases=hidbiases; topgenbiases=visbiases; save mnistpo hidtop toprecbiases topgenbiases; backpropface; Thanks for your time

  Read the article

• Week in Geek: USDA Chooses Microsoft for Cloud Services Edition

  - by Asian Angel

  This week we learned how to create geeky LED holiday lights with old bottles, dig deeper in Windows Defrag via the command prompt, use Google Chrome’s drag/drop feature to upload files easier, find great gift recommendations by looking through the How-To Geek holiday gift guide, and have fun adding Merry Christmas fonts to our computers. Photo by ntr23. Random Geek Links It has been a busy week, so we have extra news link goodness with information that is good for you to know. USDA making the move to Microsoft The U.S. Department of Agriculture has announced that it has chosen Microsoft to host things like e-mail, instant messaging, and collaboration through the software giant’s Business Productivity Online Suite. Google says it was cut off from USDA project bid Google is claiming that it was not given a chance to bid on a cloud-computing project for the U.S. Department of Agriculture, for which the contract was awarded to rival Microsoft. Apache is being forced into a Java Fork When Oracle rolled over Apache and Google’s objections to its Java plans in December, the scene was set for Apache to leave and, eventually, force a Java code fork. Tumblr explains daylong outage After experiencing an outage that started on Sunday afternoon and stretched through most of the day yesterday, Tumblr has explained what happened. Google demos Chrome OS, launches pilot program During a press briefing this week in San Francisco, Google launched the Chrome application store and demonstrated Chrome OS, its browser-centric netbook operating system. Don’t expect Spotify in U.S. this holiday season As of last week, Spotify had yet to sign a single licensing deal with a major label, after spending more than a year negotiating, multiple music sources told CNET. December 2010 Patch Tuesday will come with most bulletins ever According to the Microsoft Security Response Center, Microsoft will issue 17 Security Bulletins addressing 40 vulnerabilities on Tuesday, December 14. It will also host a webcast to address customer questions the following day. Hacker plants back door in Symbian firmware Indian hacker Atul Alex has had a look at the firmware for Symbian S60 smartphones and come up with a back door for it. PC quarantines raise tough complexities The concept of quarantining PCs to prevent widespread infection is “interesting, but difficult to implement, with far too many problems”, said security experts. Symantec: DDoS attacks hard to defend It has surfaced that the distributed denial of service (DDoS) attacks on Visa and MasterCard Web sites on Wednesday were carried out by a toolkit known as low orbit ion cannon (LOIC). Web Sockets and the risks of unfinished standards Enthusiasm for a promising new standard called Web Sockets has quickly cooled in some quarters as a potential security problem led some browser makers to hastily postpone support. Internet Explorer 9 to get tracking protection Microsoft is making changes to Internet Explorer 9’s security features that will better enable users to keep sites from tracking their activity across browsing sessions. NASA sold PCs with sensitive data NASA failed to remove sensitive data from computers that it sold, according to an audit report released this week. Cybercrooks create fake Amazon receipts The bad guys have created yet another online scam, this one involving fake Amazon receipts. World of Warcraft character move fees waived Until December 22, Blizzard will allow free realm transfers from 25 highly populated servers to alleviate log-in queues or performance issues. (The free transfers are one-way and one-time only.) SpaceX Dragon reaches orbit atop a Falcon with a fiery tail The Space Exploration Technologies corporation has become the first nongovernmental entity to put a vehicle into low Earth orbit. Geek Video of the Week If birds have wings, then why are the Angry Birds using slingshots? Photo by Dorkly Bits. Wait… Birds have Wings, Why are the Angry Ones Using Slingshots? Sysadmin Geek Tips How To Setup Email Alerts on Linux Using Gmail or SMTP Linux machines may require administrative intervention in countless ways, but without manually logging into them how would you know about it? Here’s how to setup emails to get notified when your machines want some tender love and attention. Random TinyHacker Links Red Panda Webcam Support Firefox and the Knoxville Zoo’s Red Panda program. Christmas Icons (Icons we like) Superb set of holiday icons by lgp85 at deviantArt. Download the .zip and use as .png or convert to .ico at Convertico.com or with tiny app Imagicon. Super User Questions Enjoy reading the great answers to this week’s popular questions from Super User Useful USB boot disks? DVD/CD burning .zip: is it more reliable, faster, longer lasting to burn a zip of files rather than the files as a folder? What are other ways to backup my files if I do not have an external drive? Anti virus what is the difference between these all? How can I block all Facebook elements/content? How-To Geek Weekly Article Recap Have you had a busy week between work and preparing for the holidays? Get caught up on your HTG reading with our hottest articles of the week. 20 Windows Keyboard Shortcuts You Might Not Know The 50 Best Registry Hacks that Make Windows Better LCD? LED? Plasma? The How-To Geek Guide to HDTV Technology HTG Explains: Which Linux File System Should You Choose? How to Use and Customize Google Chrome Web Apps One Year Ago on How-To Geek This week’s batch of retro geeky goodness is all about customizing Windows 7. ClassicShell Adds Classic Start Menu and Explorer Features to Windows 7 Get an Aero-Styled Classic Start Menu in Windows 7 Customize the Windows 7 Logon Screen Get the Classic Style Network Activity Indicator Back in Windows 7 How To Enable Check Boxes for Items In Windows 7 The Geek Note We would like you to join us in welcoming Jason Fitzpatrick to the writing staff here at How-To Geek. He started with us this past week, so take some time to read through his articles about the Wii, Kindle, & PlayStation 2 Peripherals and leave a friendly comment to say “Hi”! Got a great tip to share? Make sure to send it in to us at [email protected]. Photo by real00. Latest Features How-To Geek ETC The 50 Best Registry Hacks that Make Windows Better The How-To Geek Holiday Gift Guide (Geeky Stuff We Like) LCD? LED? Plasma? The How-To Geek Guide to HDTV Technology The How-To Geek Guide to Learning Photoshop, Part 8: Filters Improve Digital Photography by Calibrating Your Monitor Our Favorite Tech: What We’re Thankful For at How-To Geek Settle into Orbit with the Voyage Theme for Chrome and Iron Awesome Safari Compass Icons Set Escape from the Exploding Planet Wallpaper Move Your Tumblr Blog to WordPress Pytask is an Easy to Use To-Do List Manager for Your Ubuntu System Snowy Christmas House Personas Theme for Firefox

  Read the article

• Basics of Join Predicate Pushdown in Oracle

  - by Maria Colgan

  Happy New Year to all of our readers! We hope you all had a great holiday season. We start the new year by continuing our series on Optimizer transformations. This time it is the turn of Predicate Pushdown. I would like to thank Rafi Ahmed for the content of this blog.Normally, a view cannot be joined with an index-based nested loop (i.e., index access) join, since a view, in contrast with a base table, does not have an index defined on it. A view can only be joined with other tables using three methods: hash, nested loop, and sort-merge joins. Introduction The join predicate pushdown (JPPD) transformation allows a view to be joined with index-based nested-loop join method, which may provide a more optimal alternative. In the join predicate pushdown transformation, the view remains a separate query block, but it contains the join predicate, which is pushed down from its containing query block into the view. The view thus becomes correlated and must be evaluated for each row of the outer query block. These pushed-down join predicates, once inside the view, open up new index access paths on the base tables inside the view; this allows the view to be joined with index-based nested-loop join method, thereby enabling the optimizer to select an efficient execution plan. The join predicate pushdown transformation is not always optimal. The join predicate pushed-down view becomes correlated and it must be evaluated for each outer row; if there is a large number of outer rows, the cost of evaluating the view multiple times may make the nested-loop join suboptimal, and therefore joining the view with hash or sort-merge join method may be more efficient. The decision whether to push down join predicates into a view is determined by evaluating the costs of the outer query with and without the join predicate pushdown transformation under Oracle's cost-based query transformation framework. The join predicate pushdown transformation applies to both non-mergeable views and mergeable views and to pre-defined and inline views as well as to views generated internally by the optimizer during various transformations. The following shows the types of views on which join predicate pushdown is currently supported. UNION ALL/UNION view Outer-joined view Anti-joined view Semi-joined view DISTINCT view GROUP-BY view Examples Consider query A, which has an outer-joined view V. The view cannot be merged, as it contains two tables, and the join between these two tables must be performed before the join between the view and the outer table T4. A: SELECT T4.unique1, V.unique3 FROM T_4K T4,            (SELECT T10.unique3, T10.hundred, T10.ten             FROM T_5K T5, T_10K T10             WHERE T5.unique3 = T10.unique3) VWHERE T4.unique3 = V.hundred(+) AND       T4.ten = V.ten(+) AND       T4.thousand = 5; The following shows the non-default plan for query A generated by disabling join predicate pushdown. When query A undergoes join predicate pushdown, it yields query B. Note that query B is expressed in a non-standard SQL and shows an internal representation of the query. B: SELECT T4.unique1, V.unique3 FROM T_4K T4,           (SELECT T10.unique3, T10.hundred, T10.ten             FROM T_5K T5, T_10K T10             WHERE T5.unique3 = T10.unique3             AND T4.unique3 = V.hundred(+)             AND T4.ten = V.ten(+)) V WHERE T4.thousand = 5; The execution plan for query B is shown below. In the execution plan BX, note the keyword 'VIEW PUSHED PREDICATE' indicates that the view has undergone the join predicate pushdown transformation. The join predicates (shown here in red) have been moved into the view V; these join predicates open up index access paths thereby enabling index-based nested-loop join of the view. With join predicate pushdown, the cost of query A has come down from 62 to 32.  As mentioned earlier, the join predicate pushdown transformation is cost-based, and a join predicate pushed-down plan is selected only when it reduces the overall cost. Consider another example of a query C, which contains a view with the UNION ALL set operator.C: SELECT R.unique1, V.unique3 FROM T_5K R,            (SELECT T1.unique3, T2.unique1+T1.unique1             FROM T_5K T1, T_10K T2             WHERE T1.unique1 = T2.unique1             UNION ALL             SELECT T1.unique3, T2.unique2             FROM G_4K T1, T_10K T2             WHERE T1.unique1 = T2.unique1) V WHERE R.unique3 = V.unique3 and R.thousand < 1; The execution plan of query C is shown below. In the above, 'VIEW UNION ALL PUSHED PREDICATE' indicates that the UNION ALL view has undergone the join predicate pushdown transformation. As can be seen, here the join predicate has been replicated and pushed inside every branch of the UNION ALL view. The join predicates (shown here in red) open up index access paths thereby enabling index-based nested loop join of the view. Consider query D as an example of join predicate pushdown into a distinct view. We have the following cardinalities of the tables involved in query D: Sales (1,016,271), Customers (50,000), and Costs (787,766).  D: SELECT C.cust_last_name, C.cust_city FROM customers C,            (SELECT DISTINCT S.cust_id             FROM sales S, costs CT             WHERE S.prod_id = CT.prod_id and CT.unit_price > 70) V WHERE C.cust_state_province = 'CA' and C.cust_id = V.cust_id; The execution plan of query D is shown below. As shown in XD, when query D undergoes join predicate pushdown transformation, the expensive DISTINCT operator is removed and the join is converted into a semi-join; this is possible, since all the SELECT list items of the view participate in an equi-join with the outer tables. Under similar conditions, when a group-by view undergoes join predicate pushdown transformation, the expensive group-by operator can also be removed. With the join predicate pushdown transformation, the elapsed time of query D came down from 63 seconds to 5 seconds. Since distinct and group-by views are mergeable views, the cost-based transformation framework also compares the cost of merging the view with that of join predicate pushdown in selecting the most optimal execution plan. Summary We have tried to illustrate the basic ideas behind join predicate pushdown on different types of views by showing example queries that are quite simple. Oracle can handle far more complex queries and other types of views not shown here in the examples. Again many thanks to Rafi Ahmed for the content of this blog post.

  Read the article

< Previous Page | 117 118 119 120 121 122 123 124 125 126 127 128  | Next Page >