System administrators fall into two categories — those who already use monitoring and those who do not yet.
A Joke of Humor.
The need for monitoring arises in various ways. Some are fortunate enough to have monitoring come from the parent company. In this case, everything is simple; they’ve thought it all out for you — what, how, and with what to monitor. They’ve likely even already written the necessary manuals and explanations. Others come to this need on their own, usually initiated by the IT department. The downside is that you will have to learn from experience, running into issues and mistakes along the way. There are also upsides — you can choose any monitoring system and monitor only what is necessary, as well as come up with your own principles of responding to problems. Throughout different times, I worked in various companies, but where I was closer to monitoring, I followed the second path.
A Brief Excursion into the Past
My first 'experience' was a long time ago. It was with one local provider where I unexpectedly found myself as a stock clerk. Managed equipment was expensive back then, and therefore theft and outages were monitored using Friendly Pinger by pinging a few clients who were always or almost always online. It worked somewhat, but there was nothing better.
Later, with another local provider, the admins used Nagios. For the most part, I didn’t have access there, so I couldn’t evaluate its capabilities. However, managed equipment was used at each location, and likely, monitoring was an effective tool.
Then I joined a company that was a major provider and offered home internet as a subsidiary service. Here, Zenoss was utilized in all its glory. I didn’t delve deeply into it, but I could feel its power and utility — just the magic of regex alone is worth it... We were collecting, setting up the system, and writing regulations with thoughtful professionals in their field.
And at my next job, I found the need to learn about problems before some chief accountant reports them. Given that there was time for creative experiments, I went to see what the industry had to offer.
The agony of choice
In fact, the choice turned out to be surprisingly simple. Of course, everyone has different preferences, so my criteria and views at that time may not be suitable for you. A few systems came to mind, and I will briefly outline my thoughts regarding them.
As a Windows admin, the first thing that came to my mind was System Center in all its glory. The primary advantage is its integration into the Microsoft environment, without any hassle, just natively. The second plus is its comprehensive approach. Let's be honest, System Center is not just a pure monitoring system — it is still an infrastructure management system. However, this is also its first drawback. Deploying this monster solely for monitoring makes no sense. If you needed various backups and to deploy a million VDS... Besides, the implementation cost is discouraging because you will have to break the bank twice — first on licenses, and then on servers, where it will reside.
Next, let's turn to the past with Nagios. This system was dismissed right away, as configuring it manually through configuration files makes it unmanageable. I don’t judge people who enjoy sifting through a thousand and a half similar lines to fix a single parameter, but I certainly do not want to do that myself.
Zenoss. An excellent system! It has everything, can do everything, and is configured at a reasonable level of complexity, but it is quite resource-heavy. We didn’t have the scale to utilize nested groups, and the engine turned out to be overly demanding on resources. What for? We opted out.
Zabbix is our choice. It attracted us with its relatively low system requirements and ease of setup. In fact, it took just a few minutes to get it running. Download the image for VMWare and hit the 'start virtual machine' button. That’s it! I’ll go even further, for our needs, that 'starter image' would have been quite sufficient, even though we soon set everything up as needed.
Cacti was initially on the list, but we just didn’t get to it. What’s the point if Zabbix took off with the first push and everyone liked it right away? So I can’t say anything about Cacti.
After what has been written
The company where I implemented Zabbix has unfortunately shut down. The owner said, "I'm tired of everything, I'm closing the business," so there isn't much to say about monitoring there. We monitored servers, the internet and tunnels across all platforms and collected counters from the printers.
Then PRTG came briefly into my life. In my opinion, it works wonderfully with Windows systems, employs an interesting agent mechanism, and costs quite a bit. This is despite the rather sad ideology of access to version updates.
The company I currently work for uses Zabbix. It wasn't my choice, but I'm happy with it and fully support it. Considering the state of the monitoring system before I joined, I almost rebuilt everything from scratch. There was an understanding that "we're doing something wrong." A new server instance with Zabbix was even deployed, but there was no one to take on the task and see it through to completion. We have yet to achieve full enlightenment in monitoring, but I hope we know the direction. The process of perfecting monitoring is endless, although I have already formulated the main points for myself.
Source: habr.com
