CODEWITHSUNDEEP
nosqltime-seriesinfluxdb
MacakM
MacakM

Reputation: 1820

Timeseries NoSQL databases

I have read about various so called time series NoSQL databases. But NoSQL has only 4 types: key-value, document, columnar and graph.

For example InfluxDB does not state which NoSQL type it has, but from the documentation it seems like simple key-value store to me.

Are these time series databases only specialized databases from one of those 4 types or it is a new type of NoSQL database?

Upvotes: 1

Views: 1104

Answers (1)

Tug Grall
Tug Grall

Reputation: 3510

So to make it short, you can find both pure time series database, or engine that runs at the top of a more generic engine like Redis, Hbase, Cassandra, Elasticsearch ...

TimeSeries Databases (TSDBs) are data engines that are focusing on saving and retrieving time-based information very efficiently.

Very often since these databases will capture "events" (systems, devices/iot, applications ticks) they have support highly concurrent writes, and they usually do a lot more writes than reads.

TSDBs are storing data points within a time series, and the timestamp is usually the main index/key; allowing very efficient time range queries (give me data point from this time to this time). Data points can be multi dimensionnal, and add tags/label.

TSDBs provide mathematical operations on datapoint: SUM, DIV AVG, ... to combine data over time.

So based on these characteristics you can find databases that offer this. As you mention you can use specialized solutions like Influx DB, Druid, Prometheus; or you can find more generic databases engines that provide native time series support, or extension, let me list some of them:

Upvotes: 1

Related Questions