MySQL, MongoDB Top AWS Database Technologies

MySQL is the clear database of choice among organizations using Stackdriver to monitor their Amazon Web Services environments, but NoSQL poster-child technology MongoDB is in second place, beating out Amazon’s own SQL-based RDS and NoSQL solution DynamoDB. RDS, Amazon’s relational database-as-a-service, comes in third with pseudo-database Redis close behind.

Trailing the top database technologies are Amazon’s fairly recent NoSQL DynamoDB and Stackdriver’s personal favorite, Cassandra.

MySQL and MongoDB are the most popular databases on AWS

It should be noted that among RDS instances, 96% are MySQL, which if considered together would widen MySQL’s already significant lead considerably.

Comparing SQL to NoSQL, NoSQL is slightly more widely-used (including Redis in the NoSQL column, an admittedly controversial choice). Not by much, though – 59.1% of AWS users run a SQL database where 60% run NoSQL.

SQL and NoSQL are used roughly equally in AWS

Taking out Redis changes things immensely, as traditional relational databases suddenly make up nearly double the usage of NoSQL.

Let us know in the comments what database you prefer and why. Also, if you feel Redis shouldn’t be included, do tell! We’ll continue watching the database usage going forward to see what trends emerge on relational vs. non-relational databases and specific technology choices.

4 thoughts on “MySQL, MongoDB Top AWS Database Technologies

  • Out of curiosity, why is Stackdriver’s personal favorite DB Cassandra?

  • Do you have stats on any other NoSQL systems? Like Riak or Couchbase? Would be interested in seeing what the fall off is after Cassandra.

  • Matt,

    Great question! We evaluated many database technologies before landing on Cassandra. The reason we chose Cassandra is that it best fit our data processing needs. We primarily have a write-heavy workload that consists of time series data. Cassandra’s data model works very well when working with time series data, and one of its biggest strengths is being able to handle lots of writes.

    One of the other factors that weighed heavily was the operational cost of running Cassandra. Since Cassandra is based on a true P2P architecture, there are no special nodes and no master election. This leads to easier automation, deployment, and maintenance in AWS. As Stackdriver has grown it has been easy to add capacity while maintaining performance and reliability in our data pipeline.

    If you’re interested in learning more on how we use Cassandra in AWS we’ve put together a presentation here:


  • Not yet – the rest were pretty small, so we cut off at Cassandra. We’ll do another pull of the data in a month or so and see if any other NoSQL systems are used enough to be included.