This discussion has been locked. The information referenced herein may be inaccurate due to age, software updates, or external references.
You can no longer post new replies to this discussion. If you have a similar question you can start a new discussion in this forum.

Solarwinds adding too much load to network

I have run into issues where the polling has added loaded and makes my CPU spike on my network devices. I want to know it there is a way to get good monitoring with out adding unneeded load to my devices?

  • So every feature you turn on adds to the number of snmp oids that get polled and your device cpu has to try to answer. There are still a few OIDs that can be a bit overwhelming for network gear, usually things related to routing or neighbor topologies because some devices keep track of a lot of relationships. Just try to be aware of things ahead of time by having an understanding of how your equipment is set up and being cautious any time an OID is likely to return hundreds or into the thousands of values. All you can really do is avoid checking the box on those kinds of features when you add the device. The specific features that are problematic vary from one device to the next so it's not easy for me to come up with a prescriptive list for you, just have to work it out on the gear you own.
  •  

    I appreciate the reply alot. Unfortunately the solarwinds server was in place when I hired on and all I did was migrate it to an upgraded version of solarwinds. I have alot of different devices and technologies to manage and unfortunately I have only been in my position slightly over a year and have no formal solarwinds training. I did find a juniper router where it was checking all 800,000 routes advertised and really spiking the router cpu and turned that off but again not so easy to quickly find them all. We are smaill enough to make how the areas I cover quite broad, from the phones to the firewall, all the switches, all the routers, all the fax machine and on and on. so I guess I was hoping to find a quick answer and it sounds like there isn't one. But thanks for the reply anyway

  • A good place you could start if you wanted to try to disable some things in bulk is Settings > All Settings > NODE & GROUP MANAGEMENT - Manage Pollers, you should be able to disable the topology pollers and routing pollers and such across many devices there.  Keep in mind that turning them off will obviously prevent you from collecting that data for alerts, maps, etc.

    Always a bit of a trick to figure out the right balance of monitoring enough to see everything you want without getting overkill and over loading the hardware.  I get people asking me to set them up 10 second polling intervals and I have to laugh and break it down to them how much that would be taxing their gear in addition to the fact that I'd need tb's of database room to hold it all.

  • And from the opposite end, the monitored devices... some routers, networking cores and switches can be configured to only allow a certain percentage of the cpu be allocated to snmp processes. It's not perfect and dependent on the vendors hardware/software supporting it but something worth keeping in mind when trying to reduce the load on networking infrastructure.. in turn protecting your network availability!