topic Re: Scaling of System Manager in Active IQ Unified Manager Discussions

Scaling of System Manager

kutner — Thu, 05 Jun 2025 07:23:56 GMT

IHAC that has about 80 FAS controllers (40 dual-controller systems) that they put into System Manager. These systems are scattered around the US. They were complaining of slow response for some of the remote systems. Besides any bandwidth issues, has System manager been tested or used for large sites with 30-50 systems? I'm not sure if it's just bandwidth (they don't have problems with the snapvaults or with filerview access) or maybe something inside system manager.

Re: Scaling of System Manager

BrendonHiggins — Mon, 21 Sep 2009 10:03:40 GMT

I have emailed this thread to the chap who may know the answer.

Bren

Re: Scaling of System Manager

watan — Mon, 21 Sep 2009 13:10:17 GMT

It looks like the initial design of the product was aimed at 10-20 nodes, with the idea that customers who needed greater scaling would migrate to DFM/Operations Manager. I haven't been able fo find any official documentation with an exact figure and will see if I can get some more details on this.

Re: Scaling of System Manager

klink — Mon, 21 Sep 2009 14:08:56 GMT

In brief, nothing extraordinary was done in the design of NSM to help it scale to large numbers of controllers.

Practically speaking, management of the individual controllers should work fine even if large numbers are added to system list. However, there isn't much threading in the notifications and health monitoring, so those will tend to bog down with large numbers of systems. The actual behavior will depend on network and system latency.

> IHAC that has about 80 FAS controllers (40 dual-controller systems) that they

> put into System Manager. These systems are scattered around the US. They

> were complaining of slow response for some of the remote systems.

Are they using AD for authentication? There have been some reports of latency when using AD authentication. Minimizing the latency to AD should help.

If the latency involves health and notifications, it can be explained by what I mentioned above. If it involves active management of individual controllers, it is likely a network latency or AD authentication issue.

> It looks like the initial design of the product was aimed at 10-20 nodes, with the idea

> that customers who needed greater scaling would migrate to DFM/Operations Manager.

Yes, that was approximately the idea. NSM offers a system list, which is a substantial improvement over FilerView. We expect that customers with tens of controllers will hit a variety of scaling issues and will want ot leverage DFM/Ops Manager.

Steve

Re: Scaling of System Manager

amiller_1 — Tue, 22 Sep 2009 05:10:45 GMT

Only additional thought might just be to sniff some of the network traffic and see how much is going back and forth.