High Memory Utilization on Orion Server

More important than Nodes, how many Elements are you monitoring?

Go to Settings - All Settings - Database Details

Then scroll down for Total Elements

Depending on the version of SolarWinds you are running the recommended number could be 10k, 40k, or 100k.

https://documentation.solarwinds.com/en/success_center/orionplatform/content/orion_platform_scalability_engine_guidelines.htm#:~:text=capacity%20is%20extended.-,Scalability%20Engine%20Guidelines%20by%20Product,-The%20following%20sections

To add to this, are you running your database server instance on the same server?

That was my first thought as well, but I went the other direction. I'm glad you brought it up.

Hi Jokerfest,

The total number of elements are 12K and I am running HCO enterprise version 2025.1.1. Our database is on a different server.

Information service v3 is the process consuming most memory. 

Wow that sounds like you should be just fine. Are you using default polling times? Were you having the save issue before 2025.1.1?

I'm still on 2024.4.1 waiting for the next version to be released. My main server v3 process is only using ~1.5 GB of memory, but it only has a couple hundred elements. I'm spread across multiple servers. My busiest polling server currently has 17k elements and it's v3 process is using < 1 GB of memory right now.

Actually my Primary is having 9k elements and secondary only 3k. My Information service v3 touches 40-50GB and completely hangs the server.

Polling time is 9 minutes but we are using 2 minutes for alert evaluation. We are also using syslog on SolarWinds.

Yeah that sounds crazy. Do you have an absolute ton of web users?

"SolarWinds Information service V3 is used by websites to talk to the database. This service also controls how polling engines communicate with each other."

crazy indeed. This is not normal and isn't a trending issue that we've seen with other users. I would recommend you create a support case for it can be invested further.

We don't have too much web users, we are mainly monitoring network instances. 

As for SolarWinds support they are also not able identify much on this.

Can you please clarify "Primary is having 9k elements and secondary only 3k"?  Are you saying you have 9K elements on your primary polling engine and 3K on an additional polling engine?

Also are the web users connecting to your primary polling engine or do you have an additional web server that everyone connects to?

If both questions are 'yes', I would highly suggest setting up an additional web server and an additional polling engine.  You want to have the least interactions with the primary poller as possible.  I would also consider redirecting all SYSLOG, Window Event logs and SNMP traps to the additional poller, too.

Make sure you enable 'auto' logoff in the web settings, that can cause performance issues.

Also, ensure that all NetPaths Services are sourced from your additional poller, and if users are using SWQL Studio use the additional rather than the primary for authentication. 

Review the number of down nodes, interfaces, servers, or applications you might have.  Having a lot of 'down' items is a silent SolarWinds killer.

My last piece of advice is to look at the SW logs.  Go to the primary and open C:\ProgramData\SolarWinds\Logs\Orion & C:\ProgramData\SolarWinds\InformationService\v3.0.  Look for high number of repetitive logs or abnormally large log files.  If you find one look at the most recent log and look for errors.  This will tell you a lot about your system and it might show the root cause of the high memory usage.