Multiple NMS Cisco Prime and SolarWinds

Hi Everyone,

Just to check if anyone here has 2 Management Software running in the production network? Example Cisco Prime and SolarWinds.

Is it ok to run both Network Management Software at the same time? I am only enabling the NPM, NTA, Alerts and Syslogs in Solarwinds.

Thanks for your response!

Find more posts tagged with

cisco prime

cisco prime infrastructure

solarwinds

npm

cisco

nms

Accepted answers

All comments

cnorborg

You can if you want, there may be some duplication of polling by the two NMS's. I used to use Cisco Prime, but most of the functionality I ran into that was duplicated in the environments I was running was done better in SW. Especially configuration management.

annereavilla

Hi Craig Thank you for your response.

Using two NMS, do you think there is an additonal overhead on the network devices side (switches, routers, wlc)?

cnorborg

Of course there is! The question is whether or not it would affect the devices, these days they should be plenty powerful to handle something like that, although older devices might stumble. Things like polling intervals and such could matter, especially if you make them shorter.

If you're really worried, run a performance monitor like Solarwinds for a few weeks to get a baseline on the devices, then turn the other one on and see how much it affects it. I doubt you'll see much of a difference...

superfly

Yeah what Craig said. In the past I've just used Cisco Prime purely for the wireless and used Solarwinds for everything else. Polling is normally a low overhead but until you try it, you won't know. I personally don't see any issues with it.

cnorborg

Yes, there is. The question would be how much would it be noticed, if at all. Back in the day, the premier NMS that was out there (HP Open View) would take down networks because it was aggressively polling the routing tables of all devices it could. But, those devices had minimal CPU power, memory, etc. compared to modern devices. So, while there is definitely overhead it depends on tons of factors including

What you're polling?
- Is it simple stuff like interface counters, or is it routing tables / etc?
How often your polling
- If you're polling interfaces once every 9 minutes on 2 or 3 different NMS's, its way different than polling them every 2 minutes on 2-3.
How good / modern is your hardware?
- Polling every 2 minutes on Cisco 3850 or 9000 series switches is way better than the same on Cisco 3500 series switches.

I'm guessing polling something like a WLC, which aggregates information for tens/hundreds or even thousands of pieces of equipment has a greater effect than polling a switch. But I'd also guess a box like that has heartier hardware too, so...

So while the answer to "is there additional overhead" is definitely "yes", whether that overhead matters or is discernible highly depends on your network, meaning what equipment you have, what your polling, how often your polling and any number of other factors.

I'd recommend to stick with default polling intervals at first, ie: Solarwinds polls interfaces every 9 minutes. Get a baseline on the CPU and memory usage for those boxes and then add your second NMS and see what effect it has. Then maybe adjust your pollers if you want or need to, and see what effect it has. If you think your CPU is high on the boxes, maybe scale the polling back. Be sure to keep track of what happens and such and pay attention to what your monitoring. If I remember right, the Cisco 2950 switches notoriously reported high CPU usage whether you were monitoring them or not, like 50-75%. But, they could handle an NMS or two just fine.

annereavilla

Hi Craig Thanks for your inputs. I will try to do your recommendation.

My alternative plan is to transfer my network devices to SWinds and leave the Access points to CPI.