Weird vsa performance issue

[M.Braak]

Hi,

I’ve a weird issue with the performance of datastores on a vsa which is hosted on the same esx server.

I have two esxi 4.1u2 servers on which two vsa 9.5 servers are running. On the vsa cluster i created 1 volume. Vsa01 on the first esx node is the gateway connection for this volume. All vm's on the same node are experiencibg high latency. When for example deleting a snapshot results in a latency of 300+ ms while the other esx server has no latency at the same time. This is very odd because no physical network link is even used. Using the tool dd from an ssh shell to create a 100 file on a volume of a locally hosted volume also takes 1,5 minute. Doing the same test on a volume of a remotely hosted vsa takes 2 seconds.

Anybody got a clue what can cause this behaviour?