Extremely high CIFS Latency


IHAC that have extremely high CIFS latency( from few hundred millisec to more than 1000 millisec) seeing from the performance advisor. Please see the attached screen capture. No errors are found in th filer Any idea ?

The filer is FAS2040, running multi protocol with NFSv3, NFv4 and CIFS. DOT - 7.3.2, OM - 4.0, NMC - 3.0.

As this is the development filer, no performance issue and complaint so far.



Re: Extremely high CIFS Latency


Performance issues are often more complex than they seem at the outset.  You haven't really said much about what the filer is doing: what sort of data, access patterns, volume/aggr setups, deduplication, reallocation, etc...

Without banging the "upgrade" drum too hard, I'd jump up to .  7.3.2 had (as all ONTap software inevitably has) a few nasty bugs. seems to running well for me on well-loaded machines with all protocols enabled, except NFSv4.

I'd also suggest you try running without NFSv4 as a bit of a bigger general test.  NFSv4 is still not as mature and widely-used as it perhaps should be.

You might also want to enable logging of your domain controller connections (see 'options cifs' ) and some tracking of DNS response times, as well as performance on your kerberos server.  Lags in completing authentication can cause these spikes.