Hello,
We are trying to figure out why were are starting to see poor windows linked clone performance and we wanted see if you all can help us out. Our linked clone environment is made up of 6 BL-490c G6 each having 2 Intel Xeon E5540 2.53GHz Processors with 4 Cores each. When we look at the summary tab on the Host we see Processor Sockets 2, Cores per Socket 4, Logical Processors 16, Hyperthreading Active. and 122GB of Memory. All 6 Hosts are running ESXi 5.5 and managed by vCenter 5.5. There are no resource groups configured as well Connected via Multi-Path Fiber Channel is a NetApp FAS-2240 with 24 10K Disks configured in RAD-DP and 6 400GB SSD also configured in RAID DP, this storage configuration is setup in a flash pool. We are running View 6.0.1 with one Connection and Security Server. At any given time we are running between 50 and 65 32 Bit Windows 7 Pro Desktops loaded with Office 2013 Pro, Chrome, Firefox, Microsoft CRM 2013 Client, Java, Flash, Acrobat Reader, and Silverlight.
Based on our understand and readings 6 Hosts with 16 Logical Processors equals approximately 96 vcpu. So in theory we should be able to get up to 96 single cpu single core VMs. We know theory is the operative word here. We have been playing around with cpu/core assignment and memory numbers and cannot seem to find a new sweet spot. Currently we are set at 2 vCPU with 1 vCore each, and 4GB or Memory. When we look at the vSphere DRS and click on View Resource Distribution Chart we see the below
As the graph show we are starting to get into an area of less than 100% resource assignment most of the time more yellow and some red. Great thing is I can hover over the yellow and red system find those desktops and they are for sure experiencing poor to very poor performance. So from our point of view we are becoming processor bound or do not have enough processors to support the load. Does that make sense?
TIA,
Jim