RethinkDB hopes to resurrect itself within the Linux Foundation, as announced this week, giving it a new home and a new, developer-friendly license. It’s unclear, however, how this move resolves the fundamental issues plaguing RethinkDB: Three years too late and not different enough from MongoDB to eke out an existence. As The New Stack’s data research director Lawrence Hecht put it, “The licensing was not the core problem.”
In a roundabout transaction, the Cloud Native Computing Foundation (CNCF) picked up the RethinkDB assets for just $25,000, relicensed the code under a more developer-friendly license, then gifted the code to its parent, the Linux Foundation. What it wasn’t able to buy, however, is a resolution to the thorny problems plaguing the project, as outlined by RethinkDB co-founder Slava Akhmechet in a retrospective. Under the aegis of the Linux Foundation, RethinkDB still needs to do some soul-searching as to how to substantially differentiate itself in the market.
A MongoDB retread
Back in mid-2016, RethinkDB looked like a rising star. The company raised $12.2 million in venture capital, and saw its document database soar from #70 to #46 in the DB-Engines database popularity rankings. Nine months later, however, it’s still stuck at #46. Worse, the company that shepherded its burgeoning community was forced to shut down after eight years, unable to turn RethinkDB into a viable business.
Meanwhile, MongoDB, RethinkDB’s nemesis, kept growing in popularity and the company reportedly generated more than $100 million in revenue in 2016. (Disclosure: I used to work for MongoDB from 2012 to 2014 but sold my financial interest years ago.)
How could this be?
SEE: RethinkDB is dead, and MongoDB isn’t what killed it
After all, RethinkDB was supposed to be what MongoDB would be if only it weren’t so terrible–a beautiful blend of developer and operator excellence, as the RethinkDB team believed. Outside the RethinkDB core team, others agreed, with some claiming “RethinkDB looks like a MongoDB done right: MVCC, non blocking writes, durability by default, v8, incremental vacuum, easy sharding…” Indeed, Akhmechet went so far as to accuse MongoDB of doing most everything wrong (“It was unfathomable to us why people would choose a system that barely does the thing it’s supposed to do (store data),” he said).
And yet, they have. Over and over again.
The question is how RethinkDB managed to build a great product and yet still missed the mark(et). Former MongoDB executive Kelly Stirman answered this, telling me: “RethinkDB was fixing a problem nobody cared about.” While he said he “liked RethinkDB a lot,” and acknowledged “they were doing a lot of things right,” he also suggested, “You had to think twice why it was different.” That “think twice” translated into not nearly enough momentum to oust MongoDB at the top.
With this in mind, is there any reason to believe that Linux Foundation stewardship will position RethinkDB ahead of MongoDB and other NoSQL upstarts, including fast-growing options from Amazon, Microsoft, and Google?
Community isn’t the problem
Unfortunately, the answer is “probably not” even as RethinkDB reemerges under a community-friendly Apache license. Although, not everyone agrees. Bryan Cantrill, Joyent CTO and a member of the CNCF Technical Oversight Committee, for example, told me: “While we won’t speculate on the role of the AGPL as to the demise of RethinkDB, it was quite clear that the license (coupled with the murky ownership) presented overwhelming headwind for the project.”
But did it?
SEE: Why some of the fastest growing databases are also the most experimental
After all, as much as I love the ASLv2 (and loathe the AGPL), MongoDB seems to have done just fine despite its AGPL licensing. It also has soared in popularity despite a developer community that effectively begins and ends with MongoDB, Inc. RethinkDB licensed as ASLv2 and shepherded by the Linux Foundation is still RethinkDB and, quite frankly, not enough people are “rethinking their DB” to justify its place. In fact, there’s a distinct trend toward consolidation in database choices, a shift away from polyglot programming.
By Akhmechet’s admission, RethinkDB was three years late to the market with a product that didn’t sufficiently distinguish itself from the market leader. New ownership and a new license don’t fix these fundamental problems. Looking forward, RethinkDB needs to think beyond a shiny new license and instead determine what it needs to do to differentiate itself from more popular options like MongoDB and Apache Cassandra.
Otherwise, RethinkDB risks getting a new house, but remaining the same occupant.