Blog post

DBMS Market Transformation 2021: OSDBMS Advances

By Merv Adrian | June 06, 2022 | 0 Comments

RedisRDBMSPivotalOSSOracleopen sourceNonrelational DBMSNeo4jMySQLMongoDBMicrosoftMariaDBIBMhpeGoogleGartnerEnterpriseDBDBMSDatastaxDatabricksCouchbaseCockroachDBClouderaApache HBaseApache CassandraAmazon Web ServicesAerospike

A very frequent topic of inquiries to the Gartner DBMS teams can be stated simply as: “should I consider an open source DBMS (OSDBMS)?” Users have asked about open source in 3.6% of the team’s inquiries over the past two years. Our usual answer is “yes, if it’s commercially supported and meets your requirements after a POC that tests its ability to perform as required.”

Users are not the only source of this inquiry, though. Investors want to know if OSDBMS are viable commercially, and vendors considering new opportunities often want to think about using open source as the basis for a new offering. For different reasons, they have the same question: “is OSDBMS a commercially significant market I should be interested in?”

Is OSDBMS a market juggernaut? Seemingly not. But there is a huge amount of “hidden” money here.

Among vendors generating more than $35M in revenue in 2021, 13 primarily offer a commercial product based on an OSDBMS along with a community edition of it: Aerospike, Cloudera, CockroachDB, Couchbase, Databricks, Datastax, EDB, HPE, MariaDB, MongoDB, Neo4j, Pivotal Greenplum, and Redis. Collectively, those who offer this one DBMS represented at least $3.1B in 2021 – 3.9% of the $79.5B market.

Over half of that revenue comes from two of the vendors: Cloudera and MongoDB. Neither can realistically be considered aggressive advocates for the open source, community version of their flagship DBMS offerings, but both do offer an open source version. MongoDB Community is the source-available and free to use edition of MongoDB. CDH is Cloudera’s 100% open source platform including Apache HBase (not all Cloudera revenue is attributable to the DBMSs, but Gartner lists it there.) HPE also supports HBase, but recommends its customers use the HPE Ezmeral Data Fabric Database with the Apache HBase APIs. Few other vendors offer products based on these two, though API support is more widespread.

Another modest slice of market revenue comes from numerous other small vendors with their own versions of the Big4 favorite open source DBMSs: Cassandra, MySQL, Postgres and Redis, and others with another, less broadly known OSDBMS. We can consider them the pureplays.

But there is much more revenue that is hard to quantify. The cloud service providers (CSPs) are DBMS market behemoths and offer their own versions of OSDBMSs. AWS markets RDS for MySQL and Postgres and Elasticache for Redis and Amazon Keyspaces (for Apache Cassandra). Google has Google Cloud SQL MySQL and Postgres, and Memorystore for Redis and Datastax’s Astra (based on Cassandra). Microsoft offers Azure Database for MySQL and Postgres, AzureCache for Redis and Azure Managed Instance for Apache Cassandra. Oracle offers a Community Edition of MySQL, but does not disclose the revenue from its commercial version. Oracle has upped the ante with its high-performance Heatwave offering, now in its third release with built-in machine learning; it uses the MySQL API but is not open source. In the IBM portfolio, there is IBM Cloud Database for Datastax, MySQL, Postgres and Redis. IBM Cloud Pak for Data offers OSDBMS options as well.

Like the Enterprise versions from commercial vendors, the CSP products often add some special sauce that extends the community version – in their case, it’s cloud native features that leverage their storage engines, their control of the stack, and increasingly their ability to share governance and even semantics with their other offerings.

And, as I discussed in my earlier post about nonrelational DBMS, for OSDBMS CSP revenue may well exceed all the independent players’ revenues combined.

CSP revenue for OSDBMS may well exceed all the independent players’ revenues combined.

Overall, the OSDBMS revenue story, muddy though it is, continues to be one of steady growth, appearing in more of the DBMS landscape every year. Open source is succeeding by itself and as a component of enhanced offerings. It influence and impact will continue to grow in the years ahead as more of the data management stack is disaggregated and slices continue to be replaced by open source offerings.

Comments are closed