Aurélien Hebért
A tour of TSDB queries
For many years, we’ve been recording a lot of system and application data to gain knowledge. In the monitoring world, we use this data to try to prevent system failures or to detect some application issues, as leaks or stops. The particularity of those data is that it’s evolving with time. That kind of data is called time series or metrics. A lot of open-source tools are specialized in storing them efficiently. There are the pioneers in the field like Graphite or OpenTSDB, but today a lot of new actors are changing this world, as InfluxDB, Prometheus or even Warp10. All those databases have strengths and weaknesses: on their scalability, on their integrations, on the way we can query the data. All of them propose a query system, but even if these query systems can be seen as similar, they present a lot of disparate functionalities. Some of those query systems are way more user-friendly, when others can be harder to learn, some can be more powerful and other efficient for specific use-cases. Knowing how to query the data stored is a key-changed in a monitoring project: let’s do a quick tour of the existing methods? Tie yourself, our ride can begin! How works the data retrieval with OpenTSDB? In which use case is OpenTSDB still relevant? What brings today databases as InfluxDB or Prometheus ? What are the surprises of their queries’ languages? Why today Graphite is still hyped? This presentation will not make you a time series querying expert, yet it will gives some hints on the queries capabilities of each tools and introduce the reasons behind times series languages as WarpScript, Flux and TSL.