Where Products like Zabbix Fail

[OldExperienced&Bitter]

Zabbix is a great product. It can monitor anything competently. Still, as an IT professional with over 25 years experience and 15 as a contractor, most of my clients wind up shelling out for either Solarwinds or PRTG.

It's simple and inescapable - the paid products I mentioned lower the total cost of ownership by targeting the most common American business IT environment: Cisco routers, Cisco or HP switches, and Wintel servers. Assuming the relevant nodes have been identified, start-to-finish setup time on the paid products is less than 4 hours. These products run on Windows, not Linux, removing the added burden of paying a $90K/year Linux sysadmin.

My question is simple enough: why? Why in 2014 with America's business IT environment overwhelmingly as I described have none of the free or open-source monitoring products targeted their installs accordingly? Even Zabbix, for which I have obvious affection, still requires me to hunt down templates for something as ubiquitous as a Cisco router.

Someone please turn the light on for me. After a series of failed stagings of OpenNMS, Zenoss, Nagios, Spiceworks (groan), and Zabbix, I have to ask why none of these efforts have attempted to produce what most IT environments obviously want and are willing to pay for.

[richlv]

a) you can always run zabbix as a black box - appliance. no need to expose linux to people who are not capable of dealing with it
b) default template should discover intrfaces on cisco and monitor whole bunch of things on them.
c) as for cisco specific items, it would be great if you could upload a new one or improve existing ones at https://www.zabbix.org/wiki/Zabbix_Templates

[kloczek]

Someone please turn the light on for me. After a series of failed stagings of OpenNMS, Zenoss, Nagios, Spiceworks (groan), and Zabbix, I have to ask why none of these efforts have attempted to produce what most IT environments obviously want and are willing to pay for.

Problem with monitoring something using any of these tools is not that these tools are not doing what they used to be doing.
Exact problem with zabbix is not with TemplateWhichIExactlyNeedNOW(tm).
Problem is much more deeper. Usually on monitoring anything you must answer many times on question like "what is most important in monitored metrics for me and my colleagues?".
And still there is no tool which can answer on this automatically and probably newer will be

Zabbix only provides some kind of base platform necessary to form and answer all of such questions. From whatever set of templates you may start at the moment after some time these templates will be soon some kind of cross section of initial form of these templates, random and regular events in monitored environment, exact requirements of some people and many many more minor factors.

If state of monitored environment and all what is know about env using all data from all metrics it is painting, someone who is maintaining whole monitoring state is the painter. His knowledge and skills are all these brushes and paints. Mood in studio where everything is painted are all these faults and crashes. Sketches of the current painting are all the painter previous work experiences and finally zabbix it is only the easel ..

[dafyre]

That used to be the case...

A few of the problems that I have experienced when comparing products such as SolarWinds, Cacti, and Zabbix...

For my institution, SolarWinds is just too pricey the last time I checked, it was like $15k plus a yearly fee for a network with 50 switches, 100APs, and ~25 servers.

Cacti, for its part, works extremely well with pretty much any SNMP device that I can throw at it. Cisco, HP, and other Misc SNMP devices. Feed it the IP address and SNMP community and 99% of the time, it returned everything I needed to know and gave me pretty little graphs, and up/down alerts and everything. However, at the time, Cacti was not monitoring non-snmp devices easily, so I used Zabbix for that.

Zabbix, however, when I first started using it, didn't do SNMP at all... But when 2.0 came out, I was ecstatic about being able to use SNMP with it... It worked well enough that I no longer needed Cacti.

Now that Zabbix has added working Low-Level Discovery, I think that puts it much closer with things like SolarWinds, and eliminates the need to template EVERY device like I used to have to do. I had a special template for my 24 port switches, and another one for my 48 port switches, and yet another one for my 26 (and another one for my 50) port switches...

I'd also agree with richlv... The customers don't need to know that it's Linux. Just let them have the web interface and be happy, lol. A properly configured Zabbix installation, to me (if you use all the features) can be easily "as good" as SolarWinds (arguably, I like Zabbix better).

Oy... I didn't mean to write a small book, lol. So I'll simply leave OldExperienced with this: offer to set your customers up on Zabbix and help them configure it. Especially since it's a product that you like and recommend, and it can save them a bundle of money. Instead of needing a Linux Admin at $90k per year, they can call you.