commentBlock = $("#comment_post_23246").find('.comment-blocks'); commentBlock.find('.loading').hide(); commentBlock.find('.see-all-comments').hide(); commentBlock.html("
<\/a>
\"it_user4329<\/a>
it_user4329<\/a>Senior Manager of Engineering with 501-1,000 employees<\/span><\/div>
<\/span>Real User<\/span><\/div>
<\/i><\/div>
<\/i>Report as inappropriate<\/a><\/div><\/div><\/div>

The old-school systems produced graphs every time data was gathered. This resulted in a fast user experience displaying graphs, but it caused the number of values per second to be limited by the number of graphs per second you can produce.<\/p>\n\n

Zabbix dynamically creates the graphs on demand. This reduces the number of times it much produce a graph, pushing up the number of values per second you can capture. But as the reviewer noted above, screens and individual graphs can display slowly if they contain too many data points.<\/p>\n\n

I agree with the reviewer that many or most of the default poll rates in the templates have excessive poll frequency. In fact, they are so high as to have an impact on the machine your are polling if you have very many values you are pulling. Sometimes I think that the people that create the templates only have one machine they are monitoring, and they set the poll frequency high just to have quicker graphs appear when setting up a new zabbix server. Nothing is more boring than spending a couple hours setting up a monitoring system, only to have a bunch of graphs with single dots on them because your polling cycle for disk space is every 15 minutes. But regardless of the reason for it, I think it is irresponsible to release templates with inappropriate polling cycles.<\/p>\n\n

But back to the graphs, if you have too much data, an otherwise simple graph will take a long time to display. On a screen this gets worse because you are displaying multiple graphs. So to get the best screen display performance, reduce the polling frequency to the lowest value that still produces good graphs.<\/p>\n\n

I have been knows to produce two objects for the same item, with different polling cycles. A long polling cycle for graphs that appear on screens and public viewable pages, and faster polling cycles for detailed data collection to be used in debugging.<\/p>\n\n

I\'ve used nearly all of the network monitoring systems in the 30+ years I have been monitoring networks. Zabbix is my favorite for most applications. I do use more advanced commercial systems such as NetMRI, as the commercial systems can do things like discover all of your systems, and self configure. Commercial systems like NetMRI also do deep inspection, such as VOIP quality analysis, that Zabbix simply isn\'t designed to do.<\/p>\n\n

I can do anything with Zabbix, anything that I have time to configure. But to be fair, systems like NetMRI can be configured for very large environments in 5 or 10 minutes, out of the box. But when I want to do something special, that I create code for myself, I don\'t use systems like NetMRI, I use Zabbix. Zabbix is my favorite general purpose network monitoring system. And to be fair, Zabbix is a commercial system too, when you need it to be.<\/p>\n\n

Tools like NetMRI have a lot more power to self-configure, but that power is not free... The NetMRI quote for the hospital I worked for was \$300,000!! The commercial version of Zabbix was much lower. And with some careful work with discovery templates, you could still get some self-configuration out of Zabbix. <\/p>\n\n

Solar Winds is another commercial tool in the same space as NetMRI. Solar Winds is nice, but the performance is impacted by the fact it runs on Windows, so it takes more hardware to monitor large enterprises, but it is comfortable for the Windows geeks. I\'m not a Windows geek...;)<\/p>\n\n

George<\/p><\/div>

<\/i>Like<\/span>(1<\/span>)<\/a><\/i>Reply<\/span><\/a><\/div>
<\/div><\/div>
<\/a>
\"it_user4275<\/a>
it_user4275<\/a>Engineer at DevScope<\/span><\/div>
<\/span>Vendor<\/span><\/div>
<\/i><\/div>
<\/i>Report as inappropriate<\/a><\/div><\/div><\/div>

Over the years I\'ve been using Nagios, Zenoss, GroundWork, Cacti and other SNMP+MRTG solutions; since I\'ve found Zabbix, it\'s been my number one choice. However, it still has some issues with the Reporting capabilities that need major improvement. From a \"techs\" point of view, Zabbix provides reliable metrics and a fairly simple implementation (with some tweaks discussed on this article) but from a managements point of view, it lacks the \"beautiful graphs and reports\" that Management \"likes\"!!! I\'m not very happy with the time I\'m asked to produce reports on my systems and have to setup the same reporting parameters time and time again...<\/p><\/div>