4

I have always wondered if there are multiple people monitoring a server using Performance monitor, does this create any type of load on the system? I am not sure where the counters are stored, so if there are ten or fifteen monitors running against a server, will there be any effect?

Edit:
My experience with Performance Monitor does show that the machine running the application will have a slight to moderate performance and memory hit. The scope of this question is remotely monitoring a system, where Performance Monitor is not running on the machine being monitored. Plus, the machine being monitored may have multiple monitors accessing it from different locations. Why you ask, because there are many people in the company monitoring the health of a farm or group of servers.

I haven't really noticed any extra load on monitored systems, though I ask the question to see if anyone else has experienced a load and what the threshold may be.

1
  • 1
    Without breaking the bank, is there a centralized monitoring system which could be setup that interested parties could check for system health? The real time nature of Performance monitor is very useful, but for most health checks I have noticed them on a 15 or 20 second sample, which says a web page with this information would be just as useful without all the connected monitors.
    – Brettski
    Feb 23, 2010 at 14:53

3 Answers 3

10

You can't observe what's happening without influencing the object being observed to some degree :-)

But the effect is minimal. Otherwise you'd have had a lot of people complaining about how useless performance monitor is when practically used in the field.

Edit: The more counters you use, the more overhead it does impose (as from this link). My experiences are that the overhead is usually negligible...if perfmon in normal use can kill your system, you're running on the edge of issues anyway.

1
  • I have noticed a drop in performance if I have A LOT of counters setup and I am on the machine I am monitoring. The scope of the question was remote monitoring (sorry, should of specified this). Thank you for the response.
    – Brettski
    Feb 23, 2010 at 14:42
3

You can reduce the load by running the perfmon on a different machine than the one you're monitoring and using remote counters.

Also, it may be best to run perfmon on a single machine, rather than have multiple perfmons running against your monitored server. I've never read or experimented with seeing what multiple remote perfmons do to the monitored server, it may be negligible, but it seems odd to me to do it from multiple collectors.

3
  • This does add some overhead in network transfers, though, and the counters are still being accessed and acting on the server...so really the only load you're taking off the system is the overhead of the perfmon gui itself and the video processing to update the display of the gui :-) Feb 22, 2010 at 17:09
  • And writing to the tracelog file, and probably other things as well. It might be a minor point, but it is an MS recommendation (for best results, squeeze tube from bottom) to run Perfmon from another box than your target.
    – mfinni
    Feb 22, 2010 at 18:08
  • The only reason for the separate collectors is that there are multiple people interested in machine state.
    – Brettski
    Feb 23, 2010 at 14:38
0

Well,this depends on the methods you are monitoring the server. If you monitor the performance through bypass mode or span, the effect is minimal. If you monitor the server on itself, it will use the memory and lower the performance of it.

2
  • bypass mode, span? Could you please explain or add reference? Is it the simple act of monitoring a remote system?
    – Brettski
    Feb 23, 2010 at 14:39
  • I'm not sure what @John is talking about either. "Bypass mode" and "span" sound like networking terms (as in, switchport configs), which is not germane to this discussion.
    – mfinni
    Feb 23, 2010 at 16:37

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .