CE_MvBMemberApr 16, 2015 at 9:59 am #165173
We have a hyper-v environment running with 4 physical servers.
Server Hv1&Hv2 are exactly the same and server Hv3&Hv4 are the same. The difference between the the 4 are small and consist of some extra mhz for CPU. We have a VM with client software running which has a SQL database client connection to another VM with SQL server running.
Whenever I place the client VM and the SQL server on the same host, the speed of the SQL data transfer seems to vary from server 1&2 to server 3&4.
All servers are running Windows Server 2012 R2, ‘should be’ setup similar and are connected to the SAN likewise. Even though HV1&2 should be the ‘older’ models, they outperform the newer servers when using the client software.
During some monitoring (using the Resource Monitor) we noticed that the bottleneck might have something to do with the (TDS) transfer speed between the client VM and the SQL server. There is a noticeable speed difference(slower) on the newer server which cannot be explained.
We can acknowledge that the transfer speed between the servers should be sufficient. Directly transferring between the HV make full use of the network capabilities and such transfer speed is not seen with the client VM and the SQL server, during SQL client communication (queries, etc).
This could of course be reasonable since the full NIC speed should not be used by the SQL server, however the difference in speed between the older servers and newer server is unexplainable currently. Especially since we use the same VM for this task and move it through the HV servers to monitor the speed.
Does anybody have some thoughts or ideas on how to filter this issue?
You must be logged in to reply to this topic.