Hbase vs Cassandra vs Kafka for high resolution time series data storage

Question

Between Hbase, Cassandra and Kafka, what are the pros and cons of using either technology for for high resolution (s or even ms) time series data storage?

Answer 1

First of all - Kafka, AFAIK, is not a storage system but rather a messaging platform you can use to send the time series events to a system that will process and store that information.

Second of all - it all depends on how would you like to aggregate, query and fetch the data. As with most NoSQL databases those questions are crucial before you will even consider to move forward.

Third - there are two solutions to store time series data built on top of both HBase and Cassandra, namely OpenTSDB and KairosDB. Both pretty well known, both should be able to store events with high resolution (seconds or milliseconds). Or, alternatively, you might want to take a look at Druid, but that's a little bit different approach to the problem...

Overall, I would suggest KairosDB - it's proved to be a great performant for us storing TB of time series data.