slow Hyper V virtual machine after changing the host machine with a more powerful

Hello all:
 My scenario is as follows:
 I had a physical machine with i7 and 12GB RAM, running the TMG, DC2, and server applications. The application server runs a database SQL Anywhere SyBase see 9.2.
 Sybase customers worked well.

 Now we have a server with Xeon Intel E5620 and 20Gb. more powerful than the previous we have virtual machines on this computer, the application server is assigned a network card that is not shared with any other virtual machine.

 Application server you have 10GB, you can ask more and 6 GB have been assigned to 3 processors.

 Now users complain of the Sybase client programs sometimes display when changing them takes a second or two to display options for the dropdownlist, I've seen and it's like take to get the information from the database, once the screen loads then goes fine.

 If you run the program locally or from the application server all is well.

 My surprise is that changing to a more powerful server which allocates more memory and processors, the results seem to be worse.

 Anyone have any idea what could be happening?

 All ideas are welcome.

 thank you very much.
limmontreefreeAsked:
Who is Participating?
 
James HaywoodConnect With a Mentor Commented:
Try reinstalling the integration services. Sometimes when changing hosts the VMs benefit from this.

What is the physical server? Ifs its an Intel box there are a few more steps you will need to do.
0
 
Darius GhassemConnect With a Mentor Commented:
Make sure new server has updated network drivers installed.

I would remove network card and add an legacy network card see if this speeds up the speed. You will need to do a reboot
0
 
limmontreefreeAuthor Commented:
The 2008R2 in the host machine is update with windowsupdate.

I don't download the drivers from Intel web page.

I don't understand what do you refer with "legacy  network card" i don't  know the term...

thanks
0
Simplify Active Directory Administration

Administration of Active Directory does not have to be hard.  Too often what should be a simple task is made more difficult than it needs to be.The solution?  Hyena from SystemTools Software.  With ease-of-use as well as powerful importing and bulk updating capabilities.

 
Darius GhassemCommented:
0
 
jbvernejConnect With a Mentor Commented:
hi,

I'm not sure that your issue is a VM issue, perhaps it's a HyperV's host issue.

please update your hyperV host (at least apply SP1 of windows 2008 R2 on it) based on the current recommended update for hyperV:
try SP1 or theese updates:
Hyper-V: Update List for Windows Server 2008 R2
http://social.technet.microsoft.com/wiki/contents/articles/1349.aspx 

New CPUs or Chipsets need sometime some hotfix for hyperV's hypervisor to work ...
0
 
limmontreefreeAuthor Commented:
thanks alls:
dariusg: I can find the option legacy network card, Are you sure that it is in Windows 2008 R2 Sp1? and your are right if the problems continue i'll try to find and update for the drivers.
jbvernej: The host is a Windows 2008 R2 SP1, and the symptoms don't corresponds with my problem.
hhaywood000: yes, You are right. i do it and tomorow i'll know the results.

Thanks again , tomorrow i'll tell you.
0
 
limmontreefreeAuthor Commented:
hello all,

After the reinstall of the integration services, and do the tests, problems remain the same, I have to proveto to download a newest driver from  intel's website.

thanks
0
 
jbvernejConnect With a Mentor Commented:

ok

Don't forget to update BIOS / firmware for motherboard/NIC on your hardware and drivers in the parent partition (ie host OS).
NIC teaming: double check if NIC teaming is supported for your version
0
 
limmontreefreeAuthor Commented:
jbvernej:

I just updated the nic driver, tomorrow I´ll know if that was the solution.

thanks
0
 
limmontreefreeAuthor Commented:
hello again:

I've updated all the drivers in the server expecially the nics, but the problem persists.

Actually the delay is about 10 seconds.

Has the hyper-V any configuration about NIC that Can affect the problem.?

Thanks
0
 
James HaywoodConnect With a Mentor Commented:
There are lots of websites that talk about disabling TCP offloading on both host and guest NICs. Just do a google and see if this helps.

What is the make and model of your server?
0
 
limmontreefreeAuthor Commented:
The server is an Intel:
- MotherBoard S5500BC
- Part/Model Number: E51124-453
- FRU File ID: FRU Ver 0.02

The two nics In the Mother Board are:
- Intel Gigabit 82567LM-2
  Driver date 18/08/2008  version 10.0.2.2
  used for the second Domain Controller and the TMG LAN.
-  Intel Gigabit 82574L
  Driver date 21/06/2011  version 11.11.43.0
   used for the RemoteApp server.

and a PCI NIC DLINK added in the PCI bus.
   Driver Date 21/06/2006  version 6.1.7600.16385
   used for the Host and the TMG WAN

Thanks again.

 
0
 
James HaywoodCommented:
Two things you may want to try:

Lower the amount of RAM to 32GB or less (this was a fix for either this board or the one after) This increased our network speed in VMs from 15MBps to 85MBps!

If possible reinstall Server 2008. During the installation connect the two onboard NICs with some cat5. Once the installation is complete, install the latest drivers from intel and then remove the cat5. (This is direct from Intel)
0
 
limmontreefreeAuthor Commented:

Thaks hhaywood000:

I don't understand Where i have to lower the amount of RAM,

About reinstall the server 2008 R2 actually is very difficult.

I've change some option in the NIC's advanced features:
-. download large shipping version  2 (ipv4)  :  disable (this translate is from spanish...)
-. download large shipping version  2 (ipv6)  :  disable (this translate is from spanish...)
-. giant package (jumbo frame In english????)  : disable

thanks
"
0
 
James HaywoodCommented:
With this model of Intel Board there are issues with Hyper-V when the board contains more than 32GB. You would have to remove physical RAM from the server. I've just re-read your question and you only have 20GB so you can ignore my comment!

I understand re-installing a production host is difficult. This is a fix I received direct from Intel when having issues with some servers of this spec. It is to do with the installation of drivers during setup (undating after doesn't seem to help)
0
 
jbvernejConnect With a Mentor Commented:
- Your NIC's dispatch against VM and hyperV is strange.

In prod, you should have/need more NICs, or if you stay with 3 NIC => dispatch them like this:
#1 NIC out of  HyperV dedicated to Host's Management  (this nic is never show in any hyperV virtual network - aka vSwitch)
#2 NIC dedicated to vswitch #1 (named External/LAN Virtual Network) - uncheck the option "Allow the management operating system to share this NIC", then All VM connected to with Vswitch (DC, TMG LAN, RemoteAPP)
#3 NIC dedicated to vswitch #1 (named INTERNET/WAN Virtual Network) -  uncheck the option "Allow the management operating system to share this NIC", then connect only the TMG WAN NIC to it  (to physicaly isolate WAN/internet TMG traffic ) . This #3 NIC connected directly to your internet ISP

For better config, you should have at least 1 more NIC  (if you want to separate remoteapp traffic from the other machines
For Best config, 1 NIC for each need virtual NIC in a VM + 1 NIC dedicated for host. So here 1 NIC for host +  2 nic for TMG + 1 NIc for DC + 1 NIC for remoteapp = 5 NIC

- Any errors/warning in HyperV host Eventlogs ?
- Run an Best Practice Analyzer Tools on Hyper V Role (in the Server Manager's MMC Console) and check what
0
 
limmontreefreeAuthor Commented:
thank you very much, I know that is the best configuration, but now I have these cards and I think that the problems don't occurs for the nic's configuration, they do not have much network traffic and with old server , less powerful, these delays did not occur.

 I looked at the hyper-v log and does not say anything important.

 How can I test the performance of the network?, I have not understood what you have suggested.

 again thank you very much for your interest.
0
 
limmontreefreeAuthor Commented:
Finally, the mystery is cleared, the problem was  in the client pc  that use a ODBC pointing to old IP (when i changed the server, I moved the virtual machine from the DMZ to LAN).

 I am very grateful to everyone for  ideas and advices.

Thanks.
0
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.