Community
- Command Central
- MVP Program
- Monthly Mission
- Blogs
- Groups
- Events
- Media Vault
Products
- Observability
- Network Management
- Application Management
- IT Security
- IT Service Management
- System Management
- Database Management
Content Exchange
- SolarWinds Platform
- Server & Application Monitor
- Database Performance Analyzer
- Server Configuration Monitor
- Network Performance Monitor
- Network Configuration Manager
- SQL Sentry
- Web Help Desk
Free Tools & Trials

Sluggish response since move to v7

Since installing v7 it now seems to take a lot longer logging in, accessing maps and generally opening pages.
I thought our dual Xeon 2Gb with 2 Gb of memory would last us a lot longer than it has. The database doesn'y seem any larger than before an upgrade.

Has anybody else experienced slow down since they installed v7?

Find more posts tagged with

Accepted answers

All comments

rix

I agree
The new version runs slower for us as well.
Pretty much as you describe.

Vieregg

Yes, this is my experience too. The v7 web interface runs slower and opening maps and pages takes longer than before. When I noticed this for the first time, I backed up my database and re-installed the server from scratch. This did not help either.

Regards
Dirk

FormerMember

DITTO!!! The v7 web interface is slower and opening maps and pages takes much longer than before. The question is...will SW do anything about it soon?

tullymars

ditto. tested on 6.x, got approval to buy, installed 7 and noticed serious performance differences.

Dagwood

I'm having the same issue. I did remove my custom Maps/Drawings and went with the built-in "Live" icons and that seemed to help speed things up a bit.

I thought I was having the problem because I upgraded to the SL500 from the SL250.

BryanBecker

Same here. I actually have Orion monitoring the server its on and I see a 25-30% jump in the CPU since I loaded Version 7. I was running steady at around 20%, now I'm around 50%.

keist

Same here and of course we probably will not here anything from the tech support about this issue. is everybody running the SQL server local? If anyone runs on a remote SQL server are you experiencing the same problems?

Dan Keister

DonYonce

Can I get more information about "Sluggish Response" and higher CPU Load. Please forward your contact information (phone number, best time to contact, etc) to Development@SolarWinds.Net. Also, please include any more details if you have them.

Our testing and benchmarking of V7 showed a number of performance improvements with the Polling Engines and SQL queries. The web engine should initially take longer to load for one or two users, but scale quite well for a larger number of users (V6.4 web engine did not scale well for a large number of users). This was a development trade off between quick response and scalability.

A pre-release of V7 had a CPU issue when polling memory usage on certain Novell and Sun machines, but was resolved in the final release.

We are also considering switching from GIF to JPEG encoding for Maps which will improve web performance.

Anyhow... I'd would like to get more details on these issues. I am assigning development resources to investigate this further starting tomorrow morning.

Thanks for your feedback !!!
Don

BryanBecker

Don...I emailed you at that address. My server is very sluggish ever since the upgrade. Let me know what information you need to troubleshoot this.

Seth

I'm having the same issues too. Anybody have a work around yet?

BryanBecker

Nothing yet. Support emailed me today to gather some information like interfaces polling, how frequent I'm polling an so forth. The polling has been the same since I upgraded from v6.4 I have added some interfaces and devices...but not that many.

I now have a new problem where all my trend graphs (no matter if its interface stats, cpus, etc) have gaps in them. They show up with just bullet points at some intervals but no real trend lines.

I hope this can be fixed soon...its frustrating.

DonYonce

An update ....

The "sluggish response" issues we have seen have primarily been one of three different problems.

1) A Network Trend chart was added to a Home Page or Summary View.
The Network Trend charts were added in V7 and are VERY machine intensive since they need to scan the entire SQL table in order to generate a chart. We suggest that you do not place these on your Home Page, but instead have a link to a different view that contains them.
You can also increase the Chart Cache timeout so that these charts will be generated less frequently.

2) The Orion Server is monitoring Sun or Novel CPU and memory resources and the Polling Engine is encountering SNMP errors which causing much higher than normal CPU load on the server. This was a problem in an early pre-release, but was resolved in the final V7.0 release (released after 12/1/2003).

3) The LSASS security process is running away. The LSASS process should normally run at 0% CPU load. But.. on some dual processor machines we have seen it run as high as 60%. This problem was first identified in November 2002 and reported to Microsoft. Microsoft has aknowledged that it is an issue, has worked hard in a solution, but has not been able to come up an acceptable solution that does not break structure problems in COM+ and thousands of legacy programs that depend on COM+.
It is very easy to identify this issue. Is the LSASS process on your server running 20% to 60%? If so, then the NetPerfMon Service, SQL Server, and Web Site will be CPU starved. To correct the problem, simply shutdown the NetPerfMon Service, wait for it to stop completely and then restart. If you are more interested in the gory details of LSASS and COM+, send a note to development@solarwinds.net

We still have a couple of other leads into this issue that we are continuing to investigate.

Also, based on customer feedback to the ongoing "scablability vs performance" tradeoff structure debate, we also have a couple of structure changes that will be made to the web engine that should improve performance.

Thanks,
Don

RogSchmidt

we hadda simlar problem too. but turned out to be the lsass.exe process eating up the cpu. If you kill the Orion Sys Manager instead of shutting it down, the lsass.exe will sometimes run away. Or, if you terminate your terminal services session without shutting down the Sys Manager first, the same thing can happen. A coder described it to me as disconnecting one half of a dcom conversation. the other half doesn't know the System Manager died and keeps talking. lsass.exe is in the middle and doesn't know what to do. Its a known microaoft bug. when it happens, simply restart the NPM service.
rog

Network_Guru

Login into the customer section of the Solarwinds site and check out the new Polling Engine tuner.
It is in the "unsupported" section. Make sure you have your max polls/second set
slightly above the recommended number. This might help in the missed polls.

-=Cheers=-
NG

BryanBecker

Happy to report its back to normal. I worked with tech support and found out what the problem was. If you have lsass.exe hoggin up memory in your task manager that affects alo tof things. This is an MS issue. If you have this you need to stop and start the Orion server over. Check to see if lsass.exe is still their or not. It should be gone. My cpu has dropped back to normal 30% now. This also fixed the problem with the trends graps and missing polls. This may not help any of you but its something you should check.

keist

You should contact tech support they supplied me with a multiprocessor build that increased performance of the web site drasticaly.

Dan

rix

quote:Originally posted by Don Yonce

An update ....

1) A Network Trend chart was added to a Home Page or Summary View.
The Network Trend charts were added in V7 and are VERY machine intensive since they need to scan the entire SQL table in order to generate a chart. We suggest that you do not place these on your Home Page, but instead have a link to a different view that contains them.
You can also increase the Chart Cache timeout so that these charts will be generated less frequently.

2) The Orion Server is monitoring Sun or Novel CPU and memory resources and the Polling Engine is encountering SNMP errors which causing much higher than normal CPU load on the server. This was a problem in an early pre-release, but was resolved in the final V7.0 release (released after 12/1/2003).

3) The LSASS security process is running away. The LSASS process should normally run at 0% CPU load. But.. on some dual processor machines we have seen it run as high as 60%. This problem was first identified in November 2002 and reported to Microsoft. Microsoft has aknowledged that it is an issue, has worked hard in a solution, but has not been able to come up an acceptable solution that does not break structure problems in COM+ and thousands of legacy programs that depend on COM+.
It is very easy to identify this issue. Is the LSASS process on your server running 20% to 60%? If so, then the NetPerfMon Service, SQL Server, and Web Site will be CPU starved. To correct the problem, simply shutdown the NetPerfMon Service, wait for it to stop completely and then restart. If you are more interested in the gory details of LSASS and COM+, send a note to development@solarwinds.net

We still have a couple of other leads into this issue that we are continuing to investigate.

Thanks,
Don

The biggest performance hit we noticed is maps on pages.
If a view has a map on it, the CPU load is twice as much as a page without.

Watch Task Manager Performance CPU graph while clients are actively requesting pages with maps on them. Compare it to when the clients access standard views.

The maps we are using have 20-25 elements.
Is this considered excessive?

DonYonce

rix,
20 to 25 elements is not excessive at all. We have customers with a couple hundred elements on a single map.

But... how many levels of child maps under each map do you have ?

Can we get a copy of your maps ? (forward to Development@solarwinds.net)

I'd like to place them on a development server and analyze what's going on behind the scenes.

Thanks,
Don

BryanBecker

quote:Originally posted by keist

You should contact tech support they supplied me with a multiprocessor build that increased performance of the web site drasticaly.

Dan

I know this post was outdated but is there really a seperate build if your running a multiprocessor server versus a single processor one??

Thanks.

BB

BryanBecker

quote:Originally posted by keist

You should contact tech support they supplied me with a multiprocessor build that increased performance of the web site drasticaly.

Dan

Keist....I got a totally different response from support. They said :

"Thank you for contacting SolarWinds. Orion does not currently support multi processor machines. This is something we may add into a future version of Orion. Please let me know if you have any additional questions."

I asked directly if they have a build for a multi-processor server.

BB <-confused

mgibson

I have two pollers,
1st poller on W2k3sp1 with 4 - 3 ghz cpu's, 5.5gb ram,I monitor 579 nodes, 2280 Int, 5 Vol, polling completion of 80.75% with the poller tuner maxed 300/300 recomended 128/201..
2nd Poller on W2ksp4 with 2 - 1.2ghz cpu's,2.6gb ram, I monitor 418 nodes, 3131 Int, 0 Vol, polling completeion 97.41% with the poller tuner at 70/99 and recommended 70/124
The 1st poller is noticibly slower especially when first starting Sys Mgr and then when trying to configure alerts(initially and each time).
I am running V 7.8.4