In The DevOps 2.0 Toolkit, I argued against "traditional" monitoring tools like Nagios (https://www.nagios.org/) and Icinga (https://www.icinga.org/). Instead, we chose to use Elasticsearch for both the logs and the system metrics. In the previous chapter, I reiterated the choice for using Elasticsearch as the logging solution. Can we extend its usage by storing metrics? Yes, we can. Should we do that? Should we use it as a place to store system metrics? Are there better solutions? The biggest problem with Elasticsearch, if used as a database to store system metrics, is that it is not a time series type of database. Logs benefit greatly from Elasticsearch ability to perform free text search and store data in an unstructured way. However, for system metrics, we might take advantage of a different type of data storage. We need a time series database.
Time series databases are designed around optimized ways to store and retrieve time series...