The best way to monitor a set of systems?

[Weegee]

Hi,

I was wondering if anyone knows of an efficient way of monitoring a set of computers on a network from one main computer? More specifically, I'm looking for suggestions on only a simple set of system performance information, ie. CPU usage, RAM usage, Hard Drive Availability. The network includes both Windows and Unix O/S's on them, which I think complicates the solution. Currently, I wrote a little application that does a series of SNMP queries to gather all of this information and the app would process the info, but I found that a lot of the CPU usage was being eaten up everytime it does a query. I figured there might be a better solution, so I thought I'd ask.

Any help would be much appreciated. Thanks in advanced! :)

-Weeg

[sicarie]

Do you have a dedicated box to use for this task?

What are you looking for? Real-time alerting? Or just a daily trending/collection of this stuff?

[Weegee]

Do you have a dedicated box to use for this task?

What are you looking for? Real-time alerting? Or just a daily trending/collection of this stuff?

This will be done on a dedicated box to monitor the other systems, and I am looking for real-time alerting. Thanks!

[sicarie]

I will look around, but it might not be today - if you want to look, I'd recommend browsing http://sourceforge.net and http://freshmeat.net - Open Source repositories that are usually pretty good (that's where I'm planning to look after I google it)

I saw a shop use Cacti for this, but that is not what it is designed for. It's designed for the store and report later, trending type reports. I used to get my blackberry buzzed with what we called "everything 's all right" alerts they were so poorly configured, you knew if you hadn't heard anything that there was a problem. (Translation - don't use it unless you are desperate)

[Weegee]

Heh. Thanks for looking into it. I've been doing my own investigating as well, but I'll definitely post the latest as I come across solutions.

[sicarie]

Oh, right. I'm a Linux zealot, so I was immediately going towards FOSS, but are you wanting that, or are you looking for something that you could either have consultants set up and maintain or have some sort of support contract with a company for?

That will probably change the search pretty quickly... ;)

[Weegee]

Oh, right. I'm a Linux zealot, so I was immediately going towards FOSS, but are you wanting that, or are you looking for something that you could either have consultants set up and maintain or have some sort of support contract with a company for?

That will probably change the search pretty quickly... ;)

Probably something similar to having a consultant set up and maintain. The way I have my app right now is that it runs on the main operator system. The operator would be able to look at the app and see all of the systems are working properly via GUI LED indicators. Green means good, and red means bad health. Internally, the app would compare the system performance info with some sort of threshold, and if it exceeds that, then it'll return a "bad" response. Otherwise it's "good". But, as I mentioned before, the solution that I have eats up a lot of CPU usage, around 30-40% everytime I do those SNMP queries. If there was FOSS that does what I'm looking for, I would definitely look into it.

[sicarie]

Ah, well, then initially, I'd recommend M$' MOM/System Center Product. As it's an Microsoft tool, they are usually more than happy to point you towards some of their Gold Partners to help with install/setup and maintenance..

We're just starting to look into it ourselves, so I can't offer too much direction in that area, but so far it sounds like it would be worth your time to call them up and hear their spiel, if to nothing else than baseline other products off it...

[oberon]

If there was FOSS that does what I'm looking for, I would definitely look into it.

I have experience with setting up and using Nagios as the monitoring software for a large network of switches and servers. It is absolutely worth looking into. Nagios is open source software, and can be downloaded from sourceforge.net .

[sicarie]

I have experience with setting up and using Nagios as the monitoring software for a large network of switches and servers. It is absolutely worth looking into. Nagios is open source software, and can be downloaded from sourceforge.net .

Dang, can't believe I didn't think of that one.

Thanks, oberon!

[Weegee]

Well, sorry for the late reply, but I just recently heard of Nagios being something that might help me out with my task. Not to mention that Oberon suggested this as well. So I guess this tool is worth looking into.

During the time when I first posted this thread and now, I created a small app using C# and with the help of Net-SNMP I was able to create an app that did a series of queries for the information I was interested in. It's not a real-time application like I first envisioned, but I think it will help with my task.

Anyway, thanks for your guy's help on this!

-Weeg

[sicarie]

Another commercial option I ran across is Mercury, but that's more system profiling than any sort of uptime or stats.