User: Password:
|
|
Subscribe / Log in / New account

Zabbix: State-of-the-art network monitoring (Linux.com)

Linux.com looks at Zabbix. "Zabbix is a server-agent type of monitoring software, meaning you have a Zabbix server where all gathered data is collected, and a Zabbix agent running on each host. Note that you can have more than one Zabbix server installed, but you cannot consolidate data from multiple servers in one single central server. All Zabbix data, including configuration and performance data, is stored in a relational database -- MySQL, PostgreSQL, or Oracle -- on the server. Zabbix server can run on all Unix/Linux distributions, and Zabbix agents are available for Linux, Unix (AIX, HP-UX, Mac OS X, Solaris, FreeBSD), Netware, Windows, and network devices running SNMP v1, v2, and v3."
(Log in to post comments)

Zabbix: State-of-the-art network monitoring (Linux.com)

Posted Mar 12, 2007 20:58 UTC (Mon) by einstein (subscriber, #2052) [Link]

I'd like to know specifically where zabbix is superior to big brother. We have big brother monitoring around a thousand servers in 2 data centers, and while it is capable, it is showing it's age. I'd love to find an up-to-date, supported replacement for bb.

Zabbix: State-of-the-art network monitoring (Linux.com)

Posted Mar 12, 2007 22:44 UTC (Mon) by vmole (guest, #111) [Link]

Ditto, compared to nagios. All of these systems seem complicated to set up, and really only show their strengths and weaknesses as the system grows, so "just try them out" isn't very useful.

Zabbix: State-of-the-art network monitoring (Linux.com)

Posted Mar 12, 2007 23:19 UTC (Mon) by alvherre (subscriber, #18730) [Link]

Maybe this quote from the article is useful:
I have used BigBrother and Nagios for a long time to troubleshoot network problems, and I was happy with them -- until Zabbix came along.
Amusingly, this is the first line of the article.

Zabbix: State-of-the-art network monitoring (Linux.com)

Posted Mar 12, 2007 23:24 UTC (Mon) by vmole (guest, #111) [Link]

I actually read the article. "I like Zabbix better" is not useful, because nowhere in the article is a mention of *why* the author prefers Zabbix. I don't see anything in the article that Zabbix can do that Nagios can't, except the chart generation.

What I'd be interested in is a comment from someone who has used both for a small company LAN (say 30-40 hosts, or more), and what is good about each.

Zabbix: State-of-the-art network monitoring (Linux.com)

Posted Mar 12, 2007 23:21 UTC (Mon) by vmole (guest, #111) [Link]

In (partial) answer to my own question, from the Wikipedia entry

ZABBIX is fully configurable from its web frontend. As such, it can be considered easier to use than the popular Nagios whose configuration is done through several text files. ZABBIX also combines monitoring and trending functionality whereas Nagios focuses solely on monitoring. However, ZABBIX does not yet have the same scalability or breadth of monitoring options of its more mature cousin.

Hmmm. Yeah, the innumerable Nagios configuration files can be a pain, but on the other hand, if you have a lot of hosts with similar monitoring requiremnts, it's pretty easy to automatically generate those files. Web-based config is attractive initially, but can be painful for large systems. OTOH, Zabbix may have a easy work-around. Anyone?

Zabbix: State-of-the-art network monitoring (Linux.com)

Posted Mar 13, 2007 4:21 UTC (Tue) by raven667 (subscriber, #5198) [Link]

While I have not used Zabbix I have used OpenNMS which resolves many of the issues resulting
from systems which require manually generated configuration files. OpenNMS a network discovery
component as well as a capability scanning component that populate its database with which hosts
exist and what services are on them so you never _need_ to configure these things. You can
optionally have different groups of settings based on some attribute of the system or manually
assigned catagory using filters and packages. OpenNMS also has the benefit of supporting Nagios
plugins so you can use it more easily if you have invested in a Nagios based monitoring
infrastructure.

Zabbix: State-of-the-art network monitoring (Linux.com)

Posted Mar 13, 2007 12:11 UTC (Tue) by mbottrell (guest, #43008) [Link]

I'm a long time Nagios user, but for large deployments Nagios does struggle.
OpenNMS does address this nicely.

I've kept my hand in with Zabbix, but it doesn't have the maturity of that of Nagios to even consider it for production monitoring.

If Zabbix is to 'compete' with Nagios they'll need to look at 'converters' for the Nagios config files and also the use of their extensive plugin repository.

In the interim, I'll stay with Nagios (small scale) and OpenNMS (large scale).

Zabbix: State-of-the-art network monitoring (Linux.com)

Posted Mar 15, 2007 17:41 UTC (Thu) by jordih (guest, #44101) [Link]

Zabbix has templates.

It comes with a mid-dozen of them: a standard unix host with their system metrics, the same for a Windows host, a MySql Server, and so on.

More and better:
The third time you configure a "special" host the same way the other two, you can convert (with care) the host to a template and voilĂ !: the next time, you will make only a couple of clicks and the work will be done.

Zabbix: State-of-the-art network monitoring (Linux.com)

Posted Mar 15, 2007 17:55 UTC (Thu) by vmole (guest, #111) [Link]

If I change the template, do all the hosts using that template update? Or is the template just copied to initialize the host config?

Zabbix: State-of-the-art network monitoring (Linux.com)

Posted Mar 15, 2007 18:04 UTC (Thu) by jordih (guest, #44101) [Link]

The hosts "point-to" the template.

If you change, say, the thresold of free disk space to send a zero or one (a la Nagios plugin) on the template, it "is changed" on every host that uses it.

In fact, nothing is changed but the template...

Zabbix: State-of-the-art network monitoring (Linux.com)

Posted Mar 15, 2007 18:16 UTC (Thu) by vmole (guest, #111) [Link]

Excellent! Thanks for the info.

Zabbix: State-of-the-art network monitoring (Linux.com)

Posted Mar 13, 2007 7:56 UTC (Tue) by job (guest, #670) [Link]

I tried Zabbix the other week and I must say that I'm impressed by the web interface. But compared to Nagios (which I regard as the market leader in the free software space) I found it lacking many of the tools that the latter have. So while nice to look at, I wouldn't change.

In my opinion, Zabbix just screams of NIH. Why start your own software when it is so close to a free existing one? If Zabbix was a fork of Nagios it might be very useful, at least if it could use Nagios plugins. But no, everything is written from scratch.

I am also not convinced that the SQL approach really scales. The data is heavily hierarchical and linear in time. With this knowledge there must be smarter ways of structuring the data.

The Linux.com article completely avoids any useful information (as expected). I am not impressed by the minimal network setup in the screenshots, which gives the impression of someone not actually using it in practice.

As my own experience,Zabbix does some things better than Nagios

Posted Mar 15, 2007 16:56 UTC (Thu) by jordih (guest, #44101) [Link]

I have found very difficult to monitor multitier and clustered services with Nagios.

Nagios focuses on host and does not provide tools to manage a cluster of hosts that run just one service as a whole. Because of this host approach also is very difficult to show the state of a service that depends on a chain of hosts.

Another thing I was wandering to have was a "panel of lights and gauges" that would represent those complex services.

It is very straightforward in Zabbix. You can monitorize everything you can measure. Your own scripts are easily merged into the agent configuration (more easy than with a Nagios plugin). But the strong advantage comes when data arrives to server and you make powerful arithmetics with the monitored values. You can evaluate a clustered service or a chained service with a complete set of operators.

And then you can show them:
You can design screens as a panel of elements where each element can be a sensitive map, a trend graph, a semaphore and much more, and all of them can be nested as a tree or even a graph of related information.

It helps you to draw the structure of your network or other kind of structures (your services, with their dependencies, etc.) and their level of duty, and helps to tell them to the users that can ask for it, from the boss to the bottom level.

Information can be shown pruned depending on your access rights. And much more, I think:

I cannot be exhaustive with Zabbix because it does not have a right documentantion and I am learning some features just walking through the code and just using the software.

As a spot: With Nagios you have an annoying screen of green or red words. With Zabbix you can have a panel-map-of-lights with a short effort.

As my own experience,Zabbix does some things better than Nagios

Posted Mar 15, 2007 17:30 UTC (Thu) by vmole (guest, #111) [Link]

Thanks, that's exactly the kind of comment/review/whatever I was looking for. And you're not kidding about the lack of documentation. It looks like the Zabbix crew is trying to get paid to set up and monitor Zabbix installations, which is fine, but means the free docs are pretty minimal.


Copyright © 2007, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds