Windows Server 2012 Core

Software-based VM-centric and flash-friendly VM storage + free version

Moderators: anton (staff), art (staff), Max (staff), Anatoly (staff)

Post Reply
epalombizio
Posts: 67
Joined: Wed Oct 27, 2010 3:40 pm

Fri Apr 12, 2013 5:45 pm

Hi,
Is Windows Server 2012 Core a recommended OS or am I better off going with 2008 r2 Core for a new installation?

Has there been any documented performance benefits of Core vs. normal installations of Windows as it pertains to Starwind installations?

I'll have two units in an HA configuration.

Thanks,
Elvis
User avatar
Anatoly (staff)
Staff
Posts: 1675
Joined: Tue Mar 01, 2011 8:28 am
Contact:

Mon Apr 15, 2013 8:09 am

My personal opinion is the new one is better one, but from StarWind iSCSI SAN software - both OSes are great.
I think it will be great if you could take a closer look at our Best Practices documentation:
http://www.starwindsoftware.com/starwin ... ces-manual
Best regards,
Anatoly Vilchinsky
Global Engineering and Support Manager
www.starwind.com
av@starwind.com
epalombizio
Posts: 67
Joined: Wed Oct 27, 2010 3:40 pm

Tue Apr 16, 2013 1:32 pm

Thanks.. I'll plan on benchmarking both operating systems to see if there is any performance difference. I prefer to go with the newer OS, but I didn't know if it was vetted by Starwind.
epalombizio
Posts: 67
Joined: Wed Oct 27, 2010 3:40 pm

Wed Apr 17, 2013 8:42 pm

After running tests between Win2k8 and Win2k12 using the same jumbo packet and optimizations listed under this post: http://www.starwindsoftware.com/forums/ ... t2293.html It would seem that Win2k8 wins out ever so slightly.

Test setup:
Win2k12 VM in vSphere host
HDTunePro
Secondary VMFS volume used for testing.

Are there any other recommendations when using Starwind with Server 2012?

I usually always stick to the newest OS for shelf life, but want to make sure there is nothing that I'm missing on the 2k12 side that could improve it.

Thanks,
Elvis
User avatar
Anatoly (staff)
Staff
Posts: 1675
Joined: Tue Mar 01, 2011 8:28 am
Contact:

Fri Apr 19, 2013 10:22 am

I need to know the following information from you in order to provide a solution:
1. RAID array model, RAID level and stripe size used, caching mode of the array used to store HA images, Windows volume allocation unit size
2. NIC models, driver versions (driver manufacturer/release date) and NIC advanced settings (Jumbo Frames, iSCSI offload etc.)
3. Network scheme.
Also, there is a document for pre-production SAN benchmarking:
http://www.starwindsoftware.com/starwin ... ice-manual


Please provide me with the requested information and I will be able to further assist you in the troubleshooting
Best regards,
Anatoly Vilchinsky
Global Engineering and Support Manager
www.starwind.com
av@starwind.com
epalombizio
Posts: 67
Joined: Wed Oct 27, 2010 3:40 pm

Fri Apr 19, 2013 2:31 pm

Hi Anatoly,
I don't really have a problem, just making an observation. I'll still stick with Server 2012 since the performance delta is so small.

Some info related to my setup:
LSI MegaRaid 9271-8i. VD1 = 14 600GB 10k Velociraptors in a RAID50 setup. VD2 = 6 2TB 7.2k drives in a RAID5 setup. Cachecade 2.0 running with 2 Intel 520 240GB SSds in a RAID1 config. 64KB strip size.

Local Starwind disk allocation unit of 64KB. Test server's iscsi connected LUN also 64KB allocation size.

Nic models - Intel X540s, I350s.. latest builds and firmware. RSS enabled, chimney enabled.. All recommended changes from Starwind implemented.

For the tests, I had dual connections from the test server, to a Dell Powerconnect 6248, to the subject Starwind server. Jumbos enabled all of the way.

I'm considering taking the switches out of the equation and doing a direct connection from starwind to test server.. In your experience, have you seen that improve performance in a lightly loaded network?

Thanks,
Elvis
User avatar
Anatoly (staff)
Staff
Posts: 1675
Joined: Tue Mar 01, 2011 8:28 am
Contact:

Mon Apr 22, 2013 11:17 am

I'm considering taking the switches out of the equation and doing a direct connection from starwind to test server.. In your experience, have you seen that improve performance in a lightly loaded network?
Avoiding switch basically means that you are avoiding theoretical bottleneck. I think it is a good plan anyway.
Best regards,
Anatoly Vilchinsky
Global Engineering and Support Manager
www.starwind.com
av@starwind.com
eickst
Posts: 21
Joined: Sat Mar 09, 2013 5:25 am

Tue Apr 23, 2013 6:52 pm

epalombizio wrote:After running tests between Win2k8 and Win2k12 using the same jumbo packet and optimizations listed under this post: http://www.starwindsoftware.com/forums/ ... t2293.html It would seem that Win2k8 wins out ever so slightly.

Test setup:
Win2k12 VM in vSphere host
HDTunePro
Secondary VMFS volume used for testing.

Are there any other recommendations when using Starwind with Server 2012?

I usually always stick to the newest OS for shelf life, but want to make sure there is nothing that I'm missing on the 2k12 side that could improve it.

Thanks,
Elvis
Were you using the loopback accelerated version of the starwind software? For me on Server 2012 it beats the performance of 2008 handily.
User avatar
anton (staff)
Site Admin
Posts: 4021
Joined: Fri Jun 18, 2004 12:03 am
Location: British Virgin Islands
Contact:

Tue Apr 23, 2013 8:08 pm

It helps with vSAN (Native SAN) for Hyper-V. Version running on a bare metal does not utilize loopback connections.
Regards,
Anton Kolomyeytsev

Chief Technology Officer & Chief Architect, StarWind Software

Image
Post Reply