Native SAN 6.0.5587 Performance Issues

[btoups2]

I am having trouble with a cluster. Please see image below.

starwind.jpg (182.98 KiB) Viewed 12691 times

The test on the left is running from the raw disk. The one on the right is a HA Image.

The cluster is 2 node Windows 2012 (fully patched) with Native SAN 6.0.5587 with the following hardware. Setup per the "Configuring HA Storage for Live Migration on Windows Server 2012" guide.

Network
iSCSI - 10Gb Direct Connect with Intel x540's on both Sides
Sync - 10Gb Direct Connect with Intel x540's on both Sides
Heartbeat - 1Gb Direct Connect with Intel I350 to Broadcom BCM5709C
Management - 1Gb nic to Core Switch on each node
Hyper-V - Nic Team with 2x BCM5709C to Core Switch on each node

Disk
12 disk 1 TB SAS Raid 0

My money is on a simple config issue but I have yet to put my finger on it.

Any help would be appreciated.

[anton (staff)]

1) Do you have loopback acceleration enabled?

2) How much of the write-back cache do you use?

[btoups2]

1. Its installed, Is there anything required to "enable" it?

2. 2GB on that drive.

1) Do you have loopback acceleration enabled?

2) How much of the write-back cache do you use?

[btoups2]

And while I am thinking about it, What is Starwinds position on Cluster Aware Updating.

It seems feasible to me that the updates on the second server could finish and it reboot prior to the resync from the first server rebooting.

[anton (staff)]

1) I'll ask engineers to make sure you do have it working. What OS build do you have? Can you report what cmd says when run?

2) How much memory does the whole system has? Free? Can you provide a Task Manager screeenshot?

1. Its installed, Is there anything required to "enable" it?

2. 2GB on that drive.

1) Do you have loopback acceleration enabled?

2) How much of the write-back cache do you use?

[anton (staff)]

You're absolutely correct! Should be turned off.

And while I am thinking about it, What is Starwinds position on Cluster Aware Updating.

It seems feasible to me that the updates on the second server could finish and it reboot prior to the resync from the first server rebooting.

[btoups2]

Build 6.2.9200

Server 1

Server 1.jpg (123.25 KiB) Viewed 12622 times

Server 2

Server2.jpg (126.9 KiB) Viewed 12623 times

These servers have 128GB and 96GB respectivly. The cluster is not fully loaded yet, I still have my heavy load VMs running on stand alone until I can resolve this performance issue.

[btoups2]

Wanted to give a few more details about the hardware.

Server 1

Poweredge R720xd
2x E5-2665
128GB 1600MHz Ram
Perc H710p
2x 300GB - 6Gb SAS - 10k for OS - Raid 1 - Adaptive Read Ahead - Write Back - 64k Stripe Size
12x 1TB - 6Gb NL-SAS - 7.2k for VMs - Raid 0 - No Read Ahead - Write Back - 256k Strip Size

Server 2

Poweredge R710
2x E5-2665
128GB 1600MHz Ram

Perc H700
6x 600GB - 6Gb SAS - 15k for OS - Raid 10 - Adaptive Read Ahead - Write Back - 64k Stripe Size

LSI 9286CV-8eCC connected to Dell MD1220
12x 1TB - 6Gb NL-SAS - 7.2k for VMs - Raid 0 - No Read Ahead - Write Back - 256k Strip Size

[Anatoly (staff)]

May I ask you if you had any chance to benchmark the RAID array and the network separatelly from the SAN?
Here is our Benchmarking Guid that should be helpful in your case:
http://www.starwindsoftware.com/custome ... actice.pdf