Being an experienced Zabbix expert, Wolfgang Alper has implemented a variety of creative projects. At this year’s Summit, his presentation will reveal how “Deutsche Telekom Technik GmbH” integrated lots of Zabbix-Server instances into their existing IBM NetCool ticket system. The main goal was to reuse the entire existing infrastructure and the attached business processes to work with alarms created by Zabbix.
We have asked Wolfgang a few questions about creativity while working with Zabbix.
Are there any limitations for Zabbix, or it can work almost anywhere when you know how to configure the software?
From my experience, and I am sure I am speaking also for the entire team, there is really no limit where Zabbix can be used.
Sometimes, I compare Zabbix with the well known “Lego” bricks. In the same way, children can build nearly anything with “Lego” bricks, IT-administrator can build great monitoring solutions with Zabbix. Since Zabbix ships with all required components out of the box, one can start simple or be more advanced based on personal knowledge and experience.
To give an example: One of our clients had a very special use case. They provide a service to their customers to optimize searches for online shops and websites, similar to search- and recommendation engines used by for example Amazon. Depending on the customer size, these services are provided through a dedicated server (farm) or by a shared server, running multiple instances of that application. Each instance provides important customer-specific metrics.
The request, therefore, was to monitor these metrics per customer with Zabbix in a way, that any new server would automatically pick up the right instances and get the right metrics. With thanks to the “Lego” bricks in Zabbix like Low-Level-Discovery, Context-Macros, and API-Programming, this was fairly easy to implement.
What will be your tips on how to be creative with Zabbix?
In German, we have a saying: “Geht nicht gibt’s nicht!” I think the best translation will be something like “Everything is possible”. With respect to network monitoring, this is certainly true for Zabbix.
Most use cases can be implemented with Zabbix. Even if some very specific required feature seems to stop an implementation in the first run, one should not give up. Look if others met a similar challenge and how they mastered it. Check carefully the tools Zabbix provides or asked someone with more experience. In most cases, there is always a way, especially with the recent changes to tags and pre-processing.
Again one example: There is a customer building rubber rollers. The idea was to monitor parts of the production process. During the production process, the ovens (where the rubber rollers are heated up) write logfiles about the current process state for each rubber roller. Those log files provided different sorts of information.
To minimize the number of log items, we used dependent items with content extraction. However, not each log entry would match each dependant item, causing the item to go into an unsupported state. However, with the recent additions to pre-processing and custom error handling, this was solvable without any special scripting or programming.
Are there some challenging Zabbix-related projects or ideas you would like to implement in the future?
Yes, there are. While monitoring is really fine with Zabbix, we still need to find a way to satisfy management departments, who would like to see a sort of traffic light indication for important services in their company.
In principle, this is possible with Zabbix, but it still requires a good amount of work and creativity.
