Zenoss vs. Zabbix for desktop monitoring

[semo]

I'm trying to look at the utilization of about 20 desktops and make a decision where to upgrade or reshuffle. I think a 2 week sample of the CPU usage, memory usage, queue depth and total disk r/w bytes should be sufficient to make an informed decision.

Zenoss and Zabbix seem like the more modern performance monitoring systems (easier to set up and configure than the alternatives) but their main focus seems to be *nix servers. The workstations I'm trying to monitor are running Windows 7 x86 and I have domain admin privileges.

I plan to enable SNMP on the Windows boxes by deploying a script with sysocmgr and importing HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\SNMP from working machine.

Any suggestions?

 

[Lithium381]

I'm trying to look at the utilization of about 20 desktops and make a decision where to upgrade or reshuffle. I think a 2 week sample of the CPU usage, memory usage, queue depth and total disk r/w bytes should be sufficient to make an informed decision.

Zenoss and Zabbix seem like the more modern performance monitoring systems (easier to set up and configure than the alternatives) but their main focus seems to be *nix servers. The workstations I'm trying to monitor are running Windows 7 x86 and I have domain admin privileges.

I plan to enable SNMP on the Windows boxes by deploying a script with sysocmgr and importing HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\SNMP from working machine.

Any suggestions?

I actually played around with "Nagios" this past week at home and it was fairly easy to setup. I only monitored a Linux Server and a Windows workstation. For the Windows box i had to install a small(3mb) client to report data back to the Nagios Daemon, but it reported CPU/Mem/Diskspace and checked for an instance of "Explorer.exe" and "Notepad.exe" in the running proccesses list and would alert me if either were closed. Maybe look into that one.

 

[Khyron320]

Hmmmm

if its just snmp monitoring cacti can do it.

here is a guide to set it up on ubuntu ... only guide i could find:
http://www.howtoforge.com/cacti_xampp_ubuntu

just grab some templates:
http://www.generationip.com/documentation/general-documentation/87-best-cacti-templates-for-windows

Many people hate cacti but to me it seems the most simple snmp grapher out there... and i like simple.

also i just found out how to export graphs to a .png format so i can post it on a website.

 

[sourceninja]

I've never heard of Zabbix before. It looks very interesting...

 

[theevilsharpie]

I'm trying to look at the utilization of about 20 desktops and make a decision where to upgrade or reshuffle. I think a 2 week sample of the CPU usage, memory usage, queue depth and total disk r/w bytes should be sufficient to make an informed decision.

Zenoss and Zabbix seem like the more modern performance monitoring systems (easier to set up and configure than the alternatives) but their main focus seems to be *nix servers. The workstations I'm trying to monitor are running Windows 7 x86 and I have domain admin privileges.

I plan to enable SNMP on the Windows boxes by deploying a script with sysocmgr and importing HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\SNMP from working machine.

Any suggestions?

To my knowledge, Windows doesn't support the monitoring of performance counters via SNMP unless you install a third party SNMP "bridge" that maps the performance counters to SNMP MIBs that the network monitoring workstation can query.

My suggestion, as in the other thread you made on this topic, is to use Windows built-in performance logging functionality. It contains everything you need to achieve your goal.

 

[semo]

hmmm it's almost back to where I started. I know Nagios and Cacti are the most popular but from what I read, it seems to be that way because of legacy. I still haven't found anyone talking about what I need though (all these tools are concerned about whether a server went down or not). I need to compare and analyze performance figures of many workstations

@theevilsharpie, I get what you are saying but I have no idea how to use the builtin tools for 20+ workstations. I don't see perfmon scaling that way. Maybe I'm wrong but I can't find a single article or guide on how to monitor and graph performance of a single station over a long period, let alone 20.

To my knowledge, Windows doesn't support the monitoring of performance counters via SNMP unless you install a third party SNMP "bridge" that maps the performance counters to SNMP MIBs that the network monitoring workstation can query.
...

Zabbix uses an agent for the clients and Zenoss has a third party SNMP extension agent. Haven't tested them yet (not at work) but I'm guessing they will provide what I need.

 

[Khyron320]

Im confused how the default snmp service built into windows xp doesnt do what you want... here is what i have graphing on my windows xp bench machine (isnt on all the time):

FYI these are static graphs i just uploaded them from my internal web. These wont be updating.

It looks like if you want I/O activity you need a plugin though...

 

[semo]

Khyron320, did you obtain these from Nagios?

Anyway, I need something like this x20 over a course of 2 weeks without me visiting the stations. I want it all zero touch and automated after the initial setup and configuration (not visiting or manually remoting into the observed stations at any point)

 

[Khyron320]

Those are from cacti

weekly/monthly/yearly views are available ...

This is my personal network, everything is linux except this and the gamer pc isnt worth monitoring.

This is zero touch and quite easy to setup. Cacti uses mysql for its website and the data for all the graphs are stored in rrd format. Fill out the machine fqdn and its snmp community name. Check all the graphs you want. Done

Polling is adjusted in the website setup page and the cronjob.
mine is set to 5 mins which is why data is so blocky.
*/5 * * * * php /usr/share/cacti/poller.php > /dev/null 2>&1

I have a select set of graphs that i display on my internal servers webpage as .pngs which are exported whenever the poller is ran.

 

[semo]

I installed cactiEZ but I can only see my server's stats. Not sure how to add my desktops.

When I started this little project, I assumed that all you have to do is install an agent at the client and give it the server's IP and everything would sort itself out (that's how our inventory system OCS works). After trying out Zabbix, opsview and cacti it seems that's not the case. I can't find a guide that tells you "How to monitor performance on remote computer". They only tell you how the individual components work

 

[jae]

I went through testing all the big free monitoring solutions. I suggest checking out PandoraFMS if you're up to it.