Newsletter:8/Zenoss 5 Performance Data
Zenoss 5 Performance Data System
Perhaps the most visibly new feature in Zenoss 5 is the new graphing system. You’ll immediately notice that they are slicker and shinier than RRDtool, rendered in HTML5, and interactive to boot. That would be enough of a reason to migrate, but the real difference is invisible, hiding under the hood. Zenoss has utilized Open Source and Open Standards to better expose data that is crucial to how your IT Environment is functioning. <Need similar picture of old/new graph>
New Interactive Graphs
The new graphs are not just much easier on the eyes, they work for you. The data behind the graph can be exported to a CSV allowing for custom reporting outside of Zenoss’s already robust reporting engine. The graphs can also be resized to produce an Operations Heads Up Display featuring the most relevant performance data. Since Zenoss 5 no longer relies on rrd timestamps, and static graph output from RRDtool, timestamps in graphs can be displayed on the fly in each user’s timezone. Gone are the days of having to remember how many time zones ahead, or behind your specific datacenter your company’s Zenoss is, and that annoying daylight savings time.
What’s Under the Hood
The back end of the Zenoss 5 performance data system is a robust Big Data architecture featuring Redis, Hadoop HBase and OpenTSDB. Utilizing Big Data allows Zenoss users to deploy inexpensive commodity hardware to scale horizontally, rather than utilizing expensive SSD storage.
In Zenoss 4.2, RRDtool is I/O intensive, and particularly ill-suited for SSD. OpenTSDB uses a bit more space, replacing a C library with Redis, which provides throttling, correlation, and OpenTSDB for long term storage. This throttling also allows intelligent queuing of data to prevent the central collector from becoming overloaded. Redis/OpenTSDB requires more ram/processes, but results in a drastic reduction of I/O. While RRDtool deleted old data over time, OpenTSDB can store data indefinitely with no loss of performance. In Zenoss 4.2, any adjustment to the round robin database polling cycle (how often your resources are polled to collect performance metrics) required the creation of entirely new rrd files. This resulted in the loss of historical polling data. With the new system this can be done on the fly with no historical loss of data. This also means that a missed polling cycle, or other gap in performance data is simply interpolated. In Zenoss 4.2, performance data was stored on each collector. This resulted in longer page load times and the need for more resources on each collector. With Zenoss 5, all performance data ends up in OpenTSDB, preserving performance data even if a collector is unreachable.
It’s Easy to Set Up
Our new performance data system is so easy to set up that you don’t have to do anything. Zenoss utilizes Docker and the Zenoss Control Center to automate the deployment, management, and maintenance of a complex Big Data solution.
It Was Time for a Change
RRDtool is a venerable and much loved fixture in most datacenters. With the advent of the Cloud, on-demand VM Instances are spun up automatically in seconds, live for a few hours, and may not be back for weeks, if ever. The need to store more data, more efficiently, longer, and on cheaper hardware is very real. Zenoss 5 has solved this need with a simple, robust, and modern solution that will scale well into the future.