Help all ye defenders of zabbix

[fmtaylor2]

I have recently run into a situation that i will be needing your help with. Systems monitoring has been mandated where I work. Different functional groups came up with different solutions. For the UNIX group (my group) zabbix has been the solution of choice for a number of years. However another group (application admins) has started blowing out nagios across their environment, and it crossed the line a few days ago when they started installing nagios on servers that I already monitor with zabbix. This might have went unnoticed except one of the machines they didn't have permissions for and asked my group to install it. This brought it to my attention, and my response was "NO!, use zabbix, if there is something you need done that it doesn't already do, well, we can do that too". so a few hours and a few e-mail up and down the chain of command and guess what, a meeting is scheduled to resolve the monitoring issue. I have probably got a few days to a week tops to prepare, and there is no way I can possibly be ready. So, what I need from the zabbix faithful is all of the reasons that zabbix is the better choice over nagios. Preferably the hundreds of thousands of ways it is better. The longer the list, the better I will feel. I wish to defend the honor of the lovely lady zabbix without any fear the the wretched villain nagios will win the day.

[Kerrygeek]

I know one of the main reasons I chose it over Nagios (I demo-ed both) was that the pc techs could add/change machines with Zabbix, they don't have to know any Linux command line stuff, it's all GUI. It's much easier to use, especially for a non-admin or a new person, than Nagios.

Good luck,
Kerry

[nelsonab]

I'm not as well versed on Nagios but here are some things I have picked up which make Zabbix better.

- Centralized management of monitoring. You don't need to change the agent to monitor something new (so long as you have permissions). System.run is your friend for this.
- Ability to drill down on graphs. This to me is one of the biggest strengths of Zabbix. Let's say you want to look at what was going on with a server one week ago at 4pm? With Nagios and RRD graphing it's not so simple.
- Ability to pull the data out and do what you want with it. Since Zabbix uses a SQL back end you can create your own queries to display the data it gathers in almost any way shape or for you want.
- Zabbix has a much easier time monitoring things from a performance perspective than may other platforms. This is due to the nature of how Zabbix monitors things. You can not only tell if something is working, but how well it's working or not working.
- If your application is written in Java there is a JMX bridge for Zabbix which will allow the application to put monitoring information directly into Zabbix
- Other monitoring data can be pumped into the Zabbix server using zabbix_send, thus if your application is not written in Java you can still write external scripts to push application specific monitoing data into Zabbix.

Here are a couple of links from the forum to help
http://www.zabbix.com/forum/showthread.php?t=9082

http://www.zabbix.com/forum/showthread.php?t=1587

Please let us know if there is any way we can help you fight this battle. :-)

[Kai-Kai]

As far as I'm concerned I've done a job around it.
My job was to find a monitoring solution to replace the existing one, which was old, no more maintened and insufficient...

I've found 4 "good" solutions. In these solutions were Zabbix and Nagios (with web interface Centreon).
But the one I choosed was Zabbix !

Why ?
Because :
- our servers are running under Sun Solaris or Windows, and installing Nagios + Ndo + Centreon was really long and a bit difficult on solaris.
- some nagios plugins doesn't work, some other one need an old version of the plugin (exemple check_swap)
- if you don't use centreon or another web interface, you have to modify config file, in command line... it's not very userfriendly
- the majority of checks done by nagios are "plugins" whereas in Zabbix, the majority is build in the agent
- zabbix runs well directly out of the box after a few hours
- zabbix offers directly all interesting functionnalities : included fonctions, custom command lines / scripts, graphes, map, auto discovery, web monitoring !
- zabbix interface allow you to configure almost everything
- zabbix interface is really more sexy than nagios' one :P
- all is Zabbix, what I mean is we only have a single product : zabbix, with nagios, to get the same result, we need to use nagios, ndo, centreon, nrpe, nsc++...

Mmmm... ok, I think I've said all what I think. ^^
Sorry if my english is not really good, but I'm not an english native speaker.

Hoping it can help, don't hesitate to feedback !!!

And... may the force be with you.

[Kai-Kai]

So fmtaylor2, have you succeed in defending Zabbix ? ^^

[untergeek]

I chuckled when I read your post. I'm in the applications admin group where I work and I'm the one who is now successfully pushing the issue and forcing Zabbix onto the UNIX group (Full disclosure: I was previously a UNIX admin).

Bottom line: Zabbix rocks the socks off of Nagios.

Nagios does what it does well, but it has the feel of a project launched and maintained by a bunch of linux sysadmins who only had a vague idea what they wanted. It was subsequently added to and proceeded to get ugly.

Zabbix has a lot more going for it, even if it had the same origins (which I don't know). If I were to make a monitoring system I'd take what worked from other monitoring systems and say, "What don't these offer but I really, really want?" Most of the things I desire are to be found in Zabbix, and it is under steady development and improvement.

We've had really good results so far and we're not even fully migrated from Nagios yet. I hope your presentation goes well.