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

Increasing Performance

I have been working on a few steps to increase the performance of our Orion installation, but web performance stills seems a little sluggish. I'm not sure if that's jus the nature of IIS and heavy dependence on .NET, or if there are other things I can improve on.

Currently I have NPM, NTA and the primary (only) web server running on a single Dell machine - 2.66 GHz dual core Xeon processor, 8 GB RAM, Server 2003, no RAID, frequent backups.

I have moved the SQL server to a dedicated, single Dell machine - 2x2.1 GHz 6-core processors, 32 GB RAM, no RAID on the OS drive, DB on a 2 drive RAID 0, DB temp and log on a 2 drive RAID 0, frequent backups.

I have followed steps to make NTA DNS resolution on-demand. I have followed a few steps to increase the SQL server performance.

When accessing the Orion web server for the first time (ie. to access the login screen, default config) it can take 5-10+ seconds for it to load. Once I am in, most NPM resources load within a couple of seconds. when accessing NTA pages, even with charts limited to, for example, 3-3D pie charts, and top 5 apps, conversations and endpoints, the charts can take 5-10+ seconds to load.

I am curious if this seems normal? If so, fine, I am just trying to get as much performance as we can with the hardware we have. Would moving the primary (only) web server to a machine (VM), and taking that load away from the NPM/NTA server, make any noticeable difference? All servers have GB network between them.

Thanks in advance for any tips or advice you can offer.

Edit - forgot to mention we are monitoring ~215 nodes, 437 interfaces, 44 volumes, and collecting NetFlow/sFlow data from ~82 interfaces.

Find more posts tagged with

lun

san

Accepted answers

All comments

acherman

One thing that has helped a lot so far - I originally had my default time period for NTA charts set to the last 2 hours, I have now changed that to the last 30 minutes and the charts load much faster. Although increasing it manually for the view has the same delay effect.

MNPS_G

Is your polling server showing any serious load on the CPU/Memory (although I would expect CPU if either)? Your hardware for your SQL server is much better than the one we have here and we monitor many more nodes (~7500) and have NTA, but our load times sound a bit better in general. For your NTA, that might be about right.

The only other thing I could say that we did outside what you have done was our server guys setup some LUNs to our SAN for backups and tempDB files. Otherwise, I think your setup sounds correct. Given the number of nodes you are watching, I would expect the main page to load quickly. Do you have some report on the main page that would take some time to execute?

MarieB

Hello--

Have any of you seen the following post, which addresses performance enhancements for NTA.

Also, NTA also has a recently update FAQ, which can address additional questions that might arise.

HTH,

acherman

Is your polling server showing any serious load on the CPU/Memory (although I would expect CPU if either)? Your hardware for your SQL server is much better than the one we have here and we monitor many more nodes (~7500) and have NTA, but our load times sound a bit better in general. For your NTA, that might be about right.

Not really showing much load at all, and more on one processor than the other. Screenshot of perf monitor (CPU spike near the right is when I just accessed the web console):

The only other thing I could say that we did outside what you have done was our server guys setup some LUNs to our SAN for backups and tempDB files. Otherwise, I think your setup sounds correct. Given the number of nodes you are watching, I would expect the main page to load quickly. Do you have some report on the main page that would take some time to execute?

The only hefty thing on the main page is two network maps - both are 256 KB, so I understand those being slow to load. Even before that, though, getting the login screen takes far longer than getting pages with content on them.

Have any of you seen the following post, which addresses performance enhancements for NTA.

I have read through those points - I have configured on-demand DNS resolution, I want to capture more than just our top talkers (for troubleshooting, etc), and I do like the search by IP address function (for the same reasons).

And while my hardware apparently does not meet the recommended hardware specs (read: slow processors), but I don't think load is an issue.

savell

When Network Altas was introduced, the map performance (load times) increased for us quite substantially. I understand that one of the reasons for this was the way the maps are now cached on the local c drive of the web server (and results in a very high I/O load on this drive). I believe that development is working on the map load time performance in the next release.

Dave.