Posted on 10-24-2014 05:48 AM
First of all, great seeing some of you guys at JNUC! I got some great ideas I can't wait to try back at work.
So recently were getting more and more reports of slowness issues using SMB. I've been doing some research and found some good possible solutions, but ran into a little issue. Is there a standard method to benchmark SMB? Ideally I'd like to be able to show that whatever solution we found did in fact increase performance. I could always do a simple copy from my location to the test location - but was wondering if there is an actual benchmarking applications or command line that people use.
Also, in your experience have you noticed SMB issues getting better in Yosemite (since it uses SMB3 as default)?
Posted on 10-25-2014 03:27 AM
For me a least SMB looks a lot better under 10.10, browsing directory structures in particular seems to have improved.
But it may of course depend on whats at the other end of the wire.
If you wanted a really simple self made benchmark you could just create a script that date stamped, copied one fairly large file and date stamped again, then do the same process with a large number of smaller files. It would give you rough and ready numbers I guess.
Run the same script afterwards and see if things are better.
Posted on 10-25-2014 04:50 AM
Look - I was leaning towards thats direction. I have a script I use to replicate all our distribution points which I could modify to do it. But was curious if there was a standard benchmark used (sort of like Xbench for macs).
Yosemite uses SMB3 by default so it should be better in theory from the start. It seems that SMB2 was the one the was causing all these issues. So when I get back to our lab on monday I'll write up a script and do some testing.