2013-03-05 03:57 AM
Typically you are ok with latency below 20ms but It really depends on the applicationm the type of disk, the protcols being used. For instance, SQL Server databases run really well with latencies of 10ms or lower. Some hypervisors can support 35ms of latnecy before they start to show performance issues.
As a general tip, it is useful to keep in mind that LUN and FC/iSCSI read latency numbers include transmission time because the clock doesn’t stop ticking until the storage system receives the acknowledgement back from the host. Thus when the average transmission block size is similar for both the volume and the lUN (as it is here) and there is a significant delta between the volume read latency and the LUN read latency this is generally a solid indication that there is an issue with either the data path or the host itself. The system gets the data from disk and attempts to send it in good time, but either there is a problem with getting it to the host, a problem with the the host that delays acknowledging receipt of the data, or a problem with that receipt getting back to the system.
If you are experiencing high than 20ms latency factors, I suggest you open a support ticket with NetApp support and inform them you have a perfromance related issue. You might even have LUN Misalignment which could cause all sorts of performance issues and high latency factors.
2014-04-28 01:26 PM
Actually whenever my team is running sql query on sql server (VM Server) on that time we are getting "disk latency" message
In this case we are not getting any other problem regarding "lun access, data access, file, folder, application users are working any disturbance etc..nothing ......expect this message
What data do i need to collect to troubleshoot??? from NetApp and SQL Server vmware regarding Disk latency on VM MS SQL...which is we are getting whenever we are running SQL query on server.