We took a packet trace both on the core and the edge and saw in the packet trace oplocks being granted, broken, and sometimes not granted at all. We're running Windows 2003 with smb/cifs version 2.0 on the core and I was under the impression Microsoft had made lots on improvements in the 2.0 version that would mitigate the latency issue. Having said that, is there anything that could be done either on the server or the WAE's to remediate this?

Thanks..

Different applications behave in different ways, even different Microsoft Office applications.

What are the applications you are testing with? What are the test parameters (WAN characteristics, server and client OS versions, etc.) and use cases (how is the data accessed, are their multiple concurrent users)?

Thanks,

Zach

Zach,

we're testing with MS Excel 2007..The bandwidth is 6Mb. The server on the core is Windows 2003. I would map a drive from the PC on the edge to the file server and try to open an excel spreadsheet. I had to add the file server on the core to the connectivity directive, even though we're running the code level 4.0.19, as it could not be autodiscovered.

Can you also tell me please if oplock is granted ONLY when there're multiple users or even when there's only a single user?

thanks..

Greg,

The best thing for us to do is look at a packet capture of the transaction. There are some specific use cases with Microsoft Excel where the behaviour of the application causes the cached copy of the spreadsheet to be invalided. We can determine if this is happening from the packet capture.

Oplocks can be granted in both single user and multi-user cases. More information about the specific types of oplocks can be found in the CIFS Technical Reference here:

http://www.snia.org/tech_activities/CIFS/CIFS-TR-1p00_FINAL.pdf

Zach