Network slowness relating to GIS and Oracle

[Scott M.]

Ok, here is one that we are strugglling with to find a potential source of "slowness" (as defined by the users).

We have 4 staff that have new, Windows 7 machines connected to a Cisco 3750 Access Switch (connection is 1 GB, and connected through a Cisco 7975 phone). Connections are to an older Snap Server on the same access switch (same VLAN actually), as well as a License Server sitting one hop away (also 1 GB) that is connected to a 6513 (running Sup 720's).

I have other staff (in a separate building), but have fewer hops (direct to a Gig blade on a 6509, then 10GB to the same 6513 listed above) to the license server, but more hops (6509 -> 6513 -> 3750) the the Snap Server. The Snap server tis the location for all of their large map and shape files. These staff indicate speeds are great, and are happy with the performance. The initial 4 staff listed above that are on the same network as the Snap Server continually complain things are slow.

I actually believe it is an issue of unrealistic expectations, but I am intereted to see if ANYONE can think of a reason why this would be slow at all. I considered it was a hop count issue as they talk to the License server, but this is a one time connection request as they open their GIS Applications. There is no reason (nor is there any traffic) that consistently talks to the License server during the day.

Lastly, other users connected to the 3750 do NOT report any issues in any way with connections to other City resources.

Thansk for looking.

[JorgeM]

Network slowness issues will generally require deep analysis. It could be a port mismatch problem, server OS, disk subsytem, application issues, bad SQL queries, etc...

You may want to take some packet captures to see where the delays are. You may see retransmits, or simply delays in the response from the app server.

Have you done any performance monitoring to see if target systems have any issues that can be quickly identified? Anything in the OS logs (event viewer) that may indicate a problem?

it could be anything at this point.

[Scott M.]

Yes, all of those have been done (with the exception of installing WireShark on the desktop, whcih is happening today).

Ports have been verified, server utilizations have been checked, network trunks between swithces have been tested (all are in mostly an idle mode), and we have been stumbling around looking for anything.

Been at this position for over 2 years and this has been a complaint since I started. Odd thing is they were originally on 100 MB ports connections running XP, and they still complained. We upgrade to 1000 MB, then all was well for about a month, then the same compaints start up again. Next we upgrasde to Windows 7 (64 bit), and once again, all is great for a month or so, then the same complaints.

Like I stated intially, I think this might be more of an unrealistic expecation (considering the "changes" helped then it went back to normal???), but figured it was worth asking.

Will post if WireShark shows us anything.

Thanks.

[JorgeM]

It could be an unrealistic expectation, just as well as that there is a problem and the upgrades/trowing more bandwidth temporarily masks the issue.

It sounds like you have done quite a bit of troubleshooting already on the network side. This could be application related, misconfiguration of the data source, etc..

I have had similar experiences like this. It is easy to find these things when the problem lies in layers 1-3. Application layer issues generally require a lot more troubleshooting especially in those cases where you dont have the application expert onsite that understands the flow of the application and how it should be configured to work in your environment.

[Scott M.]

Thanks. Got Wireshark installed, and will be using Problem Steps Recorder to capture their actual "actions" and can correspond that to the Wireshark data.

The concensus now is we might have serious database issues, and that is where to look. Hoping the latest tools will help us narrow it down.

And I agree, it may simply be a Layer 8 issue!! :)

[JorgeM]

And I agree, it may simply be a Layer 8 issue!! :)

many times it is... Also, the higher you go up the OSI model, the more challenging it is to locate and resolve issues. Layer 8 is the most ellusive because Layer 8 wont acknolwedge that its the cause of the issue even when you have documented proof!! lol.