Additional Pollers - Automatic Load Disribution

[pstewart726]

Hi there.

We recently added additional pollers to our Solarwinds system and I was shocked that you have to manually allocate nodes to specific polling engines.

My feature request would be for the OPTION to have automatic load balancing of additional pollers in an upcoming release.  There should be consideration put into how the load distribution works and we need to retain the option to manually assign nodes still (because there are corner cases in my opinion for wanting to manually control certain nodes).

Thank you.

[ctmidnight]

We used to be able to balance polling engines "en-masse" from the primary, but that was removed a while back.

[pstewart726]

Wonder why they removed it then - problems with their algorithm?  I would really like to see it back - as our additional pollers get busier I don't want to be doing a juggling act.

[aaron.damyen]

I would love to have poller load-balancing as the default functionality.  And to be able to 'pin' a device to specific pollers if needed.  The majority of my devices will acknowledge requests from all of my pollers and only a few exceptions really need to be specifically assigned.

This could also open the door for pollers with different capabilities to balance themselves with different pollers doing different work.  A great options for upgrade/update paths on the pollers, load balancing, and troubleshooting (trickle drain a poller to find problem component polling).

[adatole]

My problem with this idea is that in large environments, some nodes can only belong to certain pollers (DMZ, remote site, etc) and some are BEST polled by engines closer to them (dual data center). So having automated load balancing would also require the ability to set up load balancing groups, rules, spheres, etc.

I'm OK with the idea in principal, but it would have to be fairly robust to be useful.

[pstewart726]

Hi Leon - I agree totally and should have elaborated on this a bit further.

Complete control over the pollers is maybe more of what I am after along with the ability to automate certain "pools" of the additional pollers.

[ctmidnight]

I believe the decent compromise would be to give the user the ability to logically group certain devices to follow rules for automated poller load balancing, and tag some devices as ineligible (root them to a specific poller for DMZ nodes, for instance).

[rharland2012]

I don't think the two models need to conflict with one another. As in virtualization, some boxes require host affinity, some do not.

[Deltona]

I voted for this and would like to add stacked poller load balancing as well.

The implementation needs to be done right and maybe reconsider the option to separate object polling from the nodes (Node-Bound polling)

I'd like the option of having a specific poller do the application polling, CCM polling, IPAM polling while polling the Nodes from a separate polling engine.

[sja]

What about du that as part of that FR?

I just see that going hand in hand ...

you could have 1 more AND with poller ID or poller Group right?

https://thwack.solarwinds.com/ideas/4757

[syncopix]

This would be a cool feature. As noted it needs to be an option. For us we're not too worried about where the polling is happening from so the ability to turn on "automatically distribute between pollers" would be great.

Almost like VMware's DRS "Hey, this node has loads of volumes and interfaces and the current polling engine is at 80% capacity but there are three other engines that will perform better. Let's automatically move it...". But with the ability to create rules such as stickiness ("this node must ALWAYS be on this polling engine").