EVA 4400 with VMWare Labmanager- VM Deployment issue

[Sivaramakrishna_1]

Hi,

EVA 4400 with 48*400GB 10k FC HDD with 12 BL460c Blades with VMWare ESX 4.5 U4 Running with LabManager. Customer reporing Performance issue while deploying VM. Its taking more time than compared with Other EVAs even with Lower configuration like EMC CX-310 with 1 Disk Enclosure, Dell Equilogic iSCSI with 15 HDDs etc.

EVA Perf data showing fine and no bottlenecks.


Can anybody share some inputs.

[sreekanthtm]

Hi,

Need more information....
> How much is the EVA cache?.
> How many Host Port is utilised? Is it properly distributed for load balancing?
> How many HBA's in Server?.
> How many VM's per Server?.
> How HBA's are distributed across VM?
> Zoning?

Rgds,
Sreekanth.

[Sivaramakrishna_1]

Hi,

> How much is the EVA cache?.
EVA Cache is 4GB for Controller Pair

> How many Host Port is utilised? Is it properly distributed for load balancing?
All 4 ports utilized and connected to 2 Fabrics.
> How many HBA's in Server?.
Dual Port HBA on each server distributed to 2 Fabrics
> How many VM's per Server?.
Around 160 VMs running across 12 ESX Hosts
> How HBA's are distributed across VM?
Both HBAs Zoning with EVA ports
> Zoning?
Zoning Configured as per Best Practises. One Zone per ESX servers.

More over its a Lab Setup. Customer will contineously deploy and undeploy VMs. He is only facing perf issue while deploying. Its taking around 8-12 min for complete VM deployment where as other storages taking 4-6 min.

Thanks,
Siva

[Santhosh D]

Hi\,

could you please verify if write cache is set to write back. Write Back caching is a speed process compare to Write through.

[Sivaramakrishna_1]

Hi Santosh,

We didn't changed any Cache Settings. As per my knowledge default settings is Write back only.

The main issue is that Even the case went to L3 level and they are telling that EVA is working fine.

Customer is on specific about time taking for Deplyoment only. After deployment he didn't found any perf issues.

[sreekanthtm]

Hi,

>>>Around 160 VMs running across 12 ESX Hosts
I feel this is the problem. 160 VM's means, 40 per host port. If all the host are accessing EVA (read/write) can cause performance issue. Whats the port utilisation shows in in EVA perf?.

Rgds,
Sreekanth.