Poor performance of MSSQL after P2V

I'm kinda new for ESX and had an experienced friend help me get our san/Server combo upward and running recently.  Here is what we have and what we have done.

2 x Dell R900.  4 x processors hexagonal core, 128 GB of RAM, 2 x 73 GB mirrored drives.

EqualLogic SAN with 16 x 1 TB SATA disks.  Connected via iSCSI using two QLogic in each server and 2 Force10 S25 Gigabit spend together 1000 full duplex with 9000 MTU.  The SAN is a unique RAID50 with 2 drives hot spares.  2 VMFS stores have been created at 1 to each.  The servers have been balanced through them based on the size and activity.  We also have the SAN for snapshot for servers 3 times per day (coming soon)

We have 7 servers running through both of them.

2 x Terminal servers

1 x Exchange Server

1 x database server (MSSQL)

1 x telephone system server

2 x domain controllers

Everything worked quite well overall.  Exchange seems a little slower, but our SQL application has lost about 40% on performance.  Nothing seems to be dubbed red but SQL queries are simply much more slow.  Users complain of the app freezing when others are running big reports and other.  I read up on best practices for SQL in ESX, but I find a lot of misinformation and contradiction.  Someone here went through the same problem and found a solid solution to increase performance?  Any advice would be helpful.  Also, keep in mind that I am a VM noob at this point, so all articles with instructions would be great.  Thank you.

I'll try and sum up what I believe to be the more important points to look at:

  • VSphere upgrade and use the new SCSI and ParaVirtualised NIC adapters that should significantly improve iSCSI performance.

  • Create the virtual computer from scratch - this vdisk in the BONE is aligned partition (create partition using diskpart)

  • Use SMP VM and plenty of RAM (at least 4 GB) to ensure that the multiprocessor HAL is used

  • Enable extended frames (9000) on all the components in the I/O path (JF are now fully supported in ESX 4)

  • Create LUNS on the SAN for system DB (including Tempdb), users of the DB and logs and present them using the MS initiator iSCSI in virtual machine.

  • Format the LUNS with the NTFS using a greater than the default Allocation size (try as large as 64 K)

  • Use stackable switches optimized for iSCSI if possible (Cisco 3750-G or equivalent). Create a group ether-channel with ports of each switch to global link for iSCSI in virtual machine speed (also gives fault in case of switch failure tolerance). It should be but of a second vNic in the virtual machine on a computer PortGroup virtual set to "Load balancing based on IP hash" (rising multiple in the vSwitch must be connected to the port channel Port on the switches.

  • If the SAN is supported, attend the virtual machine over several paths of I/O (to different controllers) - create a vSwitch separated with the same configuration as above - configure VMware with 2 vNIC iSCSI and MS iSCSI initiator using multiple addresses to target and use the iSCSI LUN (MPIO for failover or MCS (multiple connections for load balancing) works best when each iSCSI NIC is on a different subnet VLANS / IP)

  • Use FC drives in the SAN where possible, SAS are better next, then the SATA that should be avoided for transactional i/o applications.

  • Use batteries as much as possible - more young readers work better than the disks of greater and less

  • I had great success virtualize disk applications on VMware, especially in the more recent additions (vShere in particular) If you use as many of the recommendations that precede as poosible. IOmeter use to measure disk performance from a physical server using MS iSCSI initiator connected to the LUN. Connect the LUNS to a VMware VM and repeat tests. There should be no reason why you shouldn't be able to get results within 15% of the physical server.

    To troubleshoot VMware network speeds, download iPerf.exe and use the client on the VM and the server on a physical server. Test different combinations of frame sizes to ensure the maximum transfer speed.

    Make good grades so that you build a picture of what works and what does not work.

    I hope this helps!

    Sean

    Tags: VMware

    Similar Questions

    • Poor performance of Firefox after you apply the latest Microsoft Updates (Win 7 x 64); all solutions?

      Recently, I updated to Firefox 27 and noticed really horrible performance (Manager tasks indicates FF connecting a carrot, frequent "Firefox is not responding messages", even strikes in late development in the URL bar). It's on an Ideapad s205 (AMD E450 APU, Win 7 x 64, 8 GB RAM, SSD) who has always had pretty decent web browsing performance for a low end machine.

      At first, I blamed FF because poor performance coincided with the update to version 27, but now I think that the first cause to be a recent Microsoft Update. The reason is that my wife has a laptop with a same low-end APU (APU C60, hard drive 5400 RPM, 4 GB of RAM, Win 7 x 64), and she was getting much better performance with FF27 until I applied a backlog of some 90 + updates. Now she suffers from the same poor performance I am and she is pretty p * ssed at home.

      Interestingly, I have two WIn 7 x 64 Intel computers to office who have 27 FF and are fully patched, and performance are very good. Maybe this problem is specific to AMD APU.

      Any thoughts? I tried to turn off the backend Azure with some comments in this forum, but makes little or no difference.

      Thank you!

      Success! 32 FF seems to have solved my problems of performance. Thank you!

    • Storage after P2V performance bottleneck

      I recently P2Ved Windows 2000 advanced server running SQL 2000 to ESXi 4.1 U1. After P2V, the database server does not work as it should. Query that was running in 20-25 seconds on the source server takes more than 60 secods of execution and the application times out. KMOY and LVAD values are not very high around 3-5. Queue is showing 100% and he has another 80-100 applications pending. When I check from within the guest operating system, it shows high disk read queue length.

      When we check the performance on the source server, it has values queue disk high length thus but request does not meet the deadline. The source database server files on SAN LUN and VM, we use the original LUN as RDM copy. The disc has 42 percent file fragmentation. The database in question file has the size of about 35 GB and 168 k fragments. Can anyone suggest a way to solve this problem?

      PS: We have already tried the defrag, but OS cannot run the same DB file defragmentation when SQL server services are stopped.

      So here's what happened... during migration we chose HW version 7, but after the migration, it was 4. After the migration, no one noticed this (sad). But when we HW version improved to 7 lag time has been reduced. However, the length of the queue was very high but performance TSQL is ok now.

      So why the HW version changed to 4? Our hypothesis, it was to virtualize these servers using P2V 2 years back. But she didn't and when they drove back vmware tools were not deleted properly... we can see vmware tools service (automatic but not started) on the physical server, but the installation directory is missing. Thus, when we migrated it automatically changed hardware version 4. Weird conclusion but that's what I think. Or is that when you migrate win2k, converter automatically assigns hardware version 4?

      Any ideas anyone?

    • Very slow performance after P2V

      Hi all

      We did a P2V of a server and after his conversion, he must have very high use of the processor and extremely slow performance as a virtual machine. We did it,

      1. all cleanings after P2V.

      2. VMware tools update.

      3. an increase in the CPU, RAM.

      Nothing seems to work, any help please.

      Thank you

      Please confirm that you have removed the old equipment after P2V and all related material linked from the drivers, utilities, officials.

      1. open a Windows command prompt and type the following command:

      set devmgr_show_nonpresent_devices=1
      

      This sets an environment variable that tells the Device Manager to display all the devices.

      2. at the same command prompt, type

      devmgmt.msc
      

      You can also try to run msconfig and disable microsoft non-related services to help you identify what could be the cause of the problem.

      • On the Services tab
      • Select "Hide all Microsoft Services"
      • Click on "disable all".
      • Mark the VMware vCenter Converter Agent
      • Mark the VMware vCenter Converter Server
      • Click 'apply '.
      • Click "close".
      • Click "restart."
    • Poor performance InternetExplorer9

      Hello

      I have my InternetExplorer9 on a clean Windows7 VM thinapped and have Veery poor performance when I run the finished app.
      I tried many things to increase the speed. My normal ie9 starts in 5 seconds, until it's on google.com, my thinapped however ie9
      needs about 10 seconds, even after the restart of the 5th. What I've tried so far is:

      -J' have compressed the ie9.exe when generating

      -J' put 'AutoStartServices = 0' in the Package.ini generating

      -J' put 'NetRelaunch = 0' and 'AutoStartServices = 0' in the Package.ini.

      -J' set the sandbox sand-path %appdata%\microsoft\windows\sendto and the local path (tried both)

      -J' deleted some unnecessary user-specific folders in the project path

      -J' got only InternetExplorer.exe as entry point

      I tried the last few steps of each constellation, but no finished package is better than the others. You have any other suggestions for

      My problem? (I use the Thinapp_4.7.2 by the way)

      Kind regards

      Tobias

      It doesn't matter as long as it is in the right section:

      [Compression]
      CompressionType = None
      [Segregation]
      DirectoryIsolationMode = merged
      [BuildOptions]
      OptimizeFor = mΘmoire
      DisableTransactionRegistry = 1

      I forgot to mention it, but there was a new feature of Internet Explorer 9: hardware acceleration.

      Change IE9 to disable hardware acceleration has helped me get better performance in my VDI environments.

      See:

      http://support.Microsoft.com/kb/2528233

      Go to Internet Options, advanced, check ""use the software instead of GPU rendering rendering "."

      Recover application or use sbmerge to update your project file.


    • HP Mini 110-3530NR extremely poor performance on the Internet

      My HP Mini 110-3530NR Netbook has very poor performance on the Internet.  I use Windows 7 Starter.

      Say hello to new coppens1.  I apologize for the delay in my responses.  I have not received the normal e-mail notifications.

      When I had to go to msconfig earlier in our interactions you make sure that the Synaptics driver load with Windows?  Follow these steps to check.

      1. Click Start, type "msconfig" and press ENTER.
      2. Go to the Services tab and search for "HP Software Framework".  This should load with Windows for a lot of the HP software to work properly.
      3. Then go to the Startup tab and make sure that anything it either by "Synaptic" is loaded with Windows.
      4. Save your changes and exit msconfig.
      5. Then go to your touchpad settings and, even once, make sure that the settings are correct.
      6. Finally, restart the machine and make sure it works properly.

      I totally understand if you want to open a new thread as this covered a few topics!  Just make sure to send me a private message with the link so that I can make sure to help you here.

      I wish you a beautiful day JC!

    • SG300 switches have poor performance in layer 3?

      We have several switches SG300 Serices. We use them to route traffic to VLAN remote offices, the Internet connections and wireless access Points.

      In a remote office, we have a SG300-10 facility to route the HQ network and the remote office subnet. The SG300 is connected to the seat through fiber and contains several VLAN Tag. If I speed, tests on the fiber of the entrants Netwotk tag link I get decent performance, 80Mbs. If I switch to a networtk who is not priginating of the central administration and road SG300-10 package, I get poor performance. 15 - 20Mbs.

      I have Fireded until a new FW SG300 - 28 p v1.2.7.76. Added a HQ VLAN 101 and VLAN new, 1025. Mapped some Tagged and not marked for each ports.  Switch has been connected to the network of HQ as untagged VLAN 101.  I put a laptop on a unmarked port VLAN 101. Ran some tests, cam is back with 750-850Mbs. great.  Put the same computer laptop on a Port of 101 tag, set the NETWORK card for tag VLAN 101, same test, same speed, 750-850Mbs.

      I configured then laptop for tag VLAN 1025. Connected to the port 1025 VLAN scored. Done speed tests, the results were 15-20Mbs!

      Then I configured laptop for Untagged VLAN 1025. Connected to unagged port 1025 VLAN. Done speed tests, the results were 15-20Mbs!

      It was only the laptop and connect to the net on the SG300 - 28 p P.C. Why this device's performance is so poor when he needs to route?

      Other switches have v1.0.0.27 FW or FW v1.1.2.0. They have similar speed problems. All configured for layer 3.

      Thank you!

      Scott<>

      How are customer gateways and server configured?

      The default gateway for connecting devices should probably be that of the IVS, they are directly connected to. So if you have a server connecting to a SVI 192.168.1.100 then the gateway server should be 192.168.1.100, not a router upstream. Who can create a loop of road if the return circulation.

      -Tom
      Please mark replied messages useful

    • VPN poor Performance - Cisco RV220W and routers WRVS4400N

      Hello

      To one of our customer IPSec VPN is established between Cisco RV 220W and routers of Cisco WRVS4400N.

      Router VPN /ISP details are as below

      Location was Location B

      Details of the Internet

      ----------------------

      DOWNLOAD: 6 to 10 Mbps
      Upload: 1 to 2 Mbps

      Details of router

      ----------------------

      Cisco RV220W

      Firmware: 1.0.3.5

      IKE policy

      Encryption: 3DES
      Authentication: MD5
      Group: Group 2
      Life key: 28800 sec

      VPN strategy

      Encryption: 3DES
      Authentication: SHA - 1
      Group: 1024 bits (Group 2)
      Life key: 3 600 s
      Perfect Forward Secrecy: enabled

      Details of the Internet

      -------------------------

      DOWNLOAD: 1.35 Mbps
      Upload: 1.24 Mbps

      Details of router

      ----------------------

      Cisco WRVS4400N

      Firmware version: V2.0.1.3

      Phase 1

      Encryption: 3DES
      Authentication: MD5
      Group: 1024 bits (Group 2)
      Life key: 28800 sec

      Phase 2

      Encryption: 3DES
      Authentication: SHA - 1
      Group: 1024 bits (Group 2)
      Life key: 3 600 s
      Perfect Forward Secrecy: enabled

      From the day that VPN has been implemented, the performance was poor. Frequent disconnections sessions live to the VPN nodes and very low transfer rate was alarming to users.

      The servers in A location and users to the site B gets authenticated at the server DC level in A location

      Applications of Terminal Server remote as Quickbooks, QQ Evolution, attendance RX serve also the location has by users to the location B

      The login is your time and all applications are extremely slow.

      I tried to copy files between share data between two locations and the results are as follows

      Location A to location B-> 130 Kbps 140 Kbps

      Location location B A-> 150 Kbps to 160 Kbps

      What can be the problem for these poor performance VPN?

      -Change the encryption for the least secure OF THE /MD5 would have a significant impact because it can reduce the overload on the routers?

      -Even if both routers are routers SMB, it has really good VPN flow according to the data sheets. I couldn't find VPN flow mentioned in the WRVS4400N data sheet. One of the sons of CSC, I also noticed the VPN of WRVS4400N flow seemed really low as only about 1.6Mbps. (https://supportforums.cisco.com/thread/2107881)  Whereas RV220W router has VPN 90Mbps flow, according to the datasheet.

      So, what can be the cause of the problem and what can be fixes possible?

      Help, please!

      ANUP sisi

      Beginner to router Cisco VPN, please help

      RVS4000 was designed to work in a small office. It supports 5 VPN tunnels with a maxium of 2 Mbps flow measured in a laboratory environment. It has a processor that has a motor integrated IPS, who would deliver 20 Mbps LAN - WAN throughput when IPS is enabled.

      RV220W has been designed to operate in a slightly larger office with 25 IPsec VPN tunnels. It has a processor that has a built-in cryptographic engine able to deliver throughput 90 Mbps of IPsec. RV220W also supports 5 SSL VPN tunnels that can be used by employees and business partners for remote access.

    • Poor performance Alienware 13

      Hello, I have recently updated my Asus G55VW to 13 Alienware and he's surprisingly poor performance in some games from my old laptop. As far as I know that I put everything to max performance but it is still low in some areas.

      Old sheet

      Intel i7 @ 2.5 GHz to 3.4 GHz

      NVIDIA geforce 660 m

      8 GB ram

      Alienware 13

      Processor Intel i5 1.7 GHz to 2.7 GHz

      NVIDIA geforce 860 m

      8 GB ram

      So far, Borderlands 2 runs poor while others like Warframe can operate at max settings no problem.  Other games like naruto shippuden ultimate ninja storm 3 full bursting cannot run because it selects only the integrated graphics card and I can not change... Right now, I think that the graphics card integrated into the issue and if there is a way to disable it, which could make things much better. Not to mention that any help to increase performance would be greatly appreciated.

      JLM264

      Alienware 13

      Processor Intel i5 1.7 GHz to 2.7 GHz

      You have a 1.7 GHZ dual core CPU... Borderlands 2 recommends the use of 2.4 GHZ quad core CPU

      Your "old" laptop ASUS who had a 2.3 GHZ quad core CPU meets this requirement more than your new machine

      Warframe works very well, because any CPU dual core is sufficient and that's what you have

    • Win2008 R2 does not start after P2V

      Hello

      I am facing a problem while doing a P2V of Win2008 R2 and it does not start after P2V with error 'selection 0xc0000225 - initialization failed because a required device is not accessible' on VM (error see attachment). We use 4 standalon converter (also tried with PlateSpin Migrate 9.0.2) to virtualize Windows 2008 R2 SP1 (HP DL580 G7) in VMware ESX 3.5.0 (IBM 3850 M2) environment. Pleae help solve them if anyone faced this issue earliar. Thank you in advance.

      Rajat

      Hello René,.

      In fact it is not, there is a mistake in the release notes.

      You can try at least a target with Converter 5.0 reconfig.

      Kind regards.

      Plamen

    • disc Flash of poor performance

      Hello
      When I run $cellcli EI list physical disk, some of the Flash drive is showing a poor performance.


      alprod1st09: FLASH_5_0 1116M08EUW poor performance
      alprod1st09: there is no such thing as FLASH_5_1 1116M08EV9
      alprod1st09: FLASH_5_2 1116M08EVA poor performance
      alprod1st09: FLASH_5_3 1116M08EVC poor performance

      is it requires to call oracle for the replacement of flash drive... it is very critical?
      Kind regards
      ANI

      One of your flash cards shows also absent I would suggest contact you oracle for a glance. You are on the last cell server software? There used to be a problem that caused problems with the pci riser cards. It is fixed in the latest software drivers.

    • Android apps have poor performance

      I have several applications created for Android and all my folios seem poor performance on android devices. It takes a long time for any article to focus on the text (all sheets are built using PDF), and there is a significant for interactivity in a Folio delay, if it does not crash the app entirely. Has anyone else had problems with Android apps? I have not read a lot of useful information on them.

      All my applications are built for android with the dimensions of 2048 x 1536 to make them readable on the Nexus-10. All applications have been rebuilt with v27 and PDFS.

      I'd appreciate comments you can give.

      Thank you

      One thing that can help with the Active Viewer is to design to a smaller size (1024 x 768, for example), use a PDF object and the viewer intensify folios to fit larger screens. Put scale up to adjustment is supported in our 28 Viewer version of AIR and it makes a big difference in performance while maintaining the reasonable quality of the assets.

      Neil

    • Poor performance after upgrade

      Adobe... why you posted an upgrade this bad?

      Everything worked very well!

      Now it is frozen at import, the compilation is slow, time display stops and I have to wait for it to catch up

      WHY... WHY... WHY?

      Title has been changed by Kevin Monahan

      Have you disabled it?

    • SQLDev poor performance after upgrade to DB 11 g

      I have a database of 10 large short g and Developer SQL works well with. But I cloned deployed this DB at 4 new hosts and DBUA to upgrade to 11.1.0.7... and now Developer SQL runs horribly on all four. It takes 2-3 minutes to return a list of 'other users' 11 g vs 10g. The plan to explain the SQL shows 100% CPU usage, but with a cost of 38, which confuses me. Here are the Sun T5240 systems with ample CPU and memory to spare.

      I've updated the statistics of dictionary with no effect...

      Someone has ideas... is there some kind of SQL Developer framework I'm missing or something like that?

      Solution: Look for sqldeveloper.jar and extract schema.xml file. Edit the file to replace the two versions of the query as follows

      Select
      * from (SELECT username,
      10 OBJECT_COUNT - wrong County
      OF SYS. ALL_USERS to the
      WHERE AU. USER NAME! = USER)

      Replace the xml file in the jar and the problem should go away.

      P.S. A customer problem that reported that it depends on the user having privilege "select any table". It is so in your case?

    • Poor performance when multitasking on T510 Core i5, Windows 7 64, 4 GB of RAM, NVidia Optimus 512 MB

      I am sadly disappointed with my new Lenovo performance and do not know what could be the cause.

      I just got a new T510 with Core i5, 4 GB of RAM and Windows 7 64 processor. When I run a few programs, such as Photoshop, Virtualbox (850 MB of RAM allocated), Google Chrome with a few tabs I'm stuck.

      It seems that 4 GB of memory is not enough for the work that I did on a 2GB machine a week ago.

      Response time of the computer is a couple of seconds for the window switching. New programs begin to count in a few minutes. Consumption of RAM reaches 98%, even if no single program is listed to consume more than 150 MB in the Manager of tasks or resource monitor. The effects Aero cool work, however, there no sloppiness in the graphical layer of the system.

      Even now, after killing of Virtualbox, remaining processes are up to 89% of the RAM in the resource manager.

      I'm working on the power running with power setting 'Performance', but I can hardly call it a performance.

      It's ridiculous, it was supposed to be a fast multitasking machine and turns out a load of scrap metal. I used to work on 2GB, 1.6 GHz Celeron with processor Intel GMA, had Ubuntu host and Guest Windows XP with 750MB of RAM allocated, plus a couple of tools for programming and never had such problems _on ordinary basis_.

      I plan to run a Ubuntu on this machine because I can't just work this way on Windows 7... I can't even kill a process easily because it takes forever to complete a process.

      A friend of mine runs similar configuration on a slower machine T410 with CPU and the poor graphics and he says it's really strange, because he never had this kind of problem. My PC is new, 4 days from arrival, its is like 2 months old.

      Everyone knows about the similar problem? What could be the cause?

      At this point, I have Chrome with about 10 tabs, Firefox empty, empty Photoshop, Pidgin and 3 Windows Explorer running, Adobe Reader with 1 paper and Sticky Notes. Consumption of RAM is 3495MB used (Yes), 113 and 276 MB free Eve.

      What's not here?

      Please see the attached screenshot:

      http://img715.imageshack.us/i/captureybh.PNG/

      Well, the problem is resolved. After removing delivered Norton Internet Security suite, run as plate gold.

      I'll take another antivirus software, then.

      Now, even with the operation of Virtualbox, I still like 2 GB of free RAM. And no lag, no slowdowns. Woohoo.

      * beep * you Norton!

    Maybe you are looking for

    • How do I change the date on a lot of photos with the built-in date wrong

      My camera date was wrong for about a month (a few hundred photos and a handful of 'events').  I know how to adjust the date for an image both in Photos Version 1.5 (370.42.0), but I'd like to change for a lot of photos at once.  Can is it possible, y

    • Re: Metallic sound comes from my Satellite C50-A-1HF

      Hi, today I bought the Toshiba Satellite C50-A-1HF and when I tried to listen to a YouTube video, I realized that the sound is a little metal, and there is no bass. The laptop speakers are not so good, so I plugged my headphones... same problem. I pl

    • HP device manager 4.6 SP4 device mass rename

      Hello I want to use HPDM 4.6 to rename all our lightweight client devices (t620). I have a list of MAC addresses and the names I want to assign to the devices. Can I rename a single device, I can rename all devices HP-invented names, but I don't see

    • do I need to buy a charger?

      I recently got a rocket as a gift. It came with a USB cable. It will automatically cool if connected to my PC (XP)? Or I have to go out and get a charger?

    • My computer won't go to sleep.

      My computer not responding or sleep command