By: +David Herron; Date: May 30, 2017
Time series is always ordered by a time-stamp, meaning that each measurement occurred at a specific time and that the measurements are sorted by their time of occurrence.
An irregular time series can be converted to a regular time series by summarizing. One summarizes a time series, producing a second time series -- for example if the original time series was stock trades over the day, a summarized time series might be the average for each 10-minute interval over the day.
- Cassandra "Distributed columnar database" that can be used for time-series data storage. Have to implement things a regular time-series database gives for free. Ditto for MongoDB, REDIS, and Elastic
- kx -- kx systems makes "KDB" database. Popular in Financial industry. Limitation of being an in-memory database and therefore cannot share data across machines.
- OpenTSDB Strictly speaking not a database, since it is a layer on top of HBASE.
- graphite A "legacy" system that's more for metrics and monitoring, rather than a generic time-series database.
- Prometheus Very strong in the DevOps world. Not very scalable.