No one does open source better than Google. While Microsoft may have more employees contributing to open source, Google projects like Kubernetes and TensorFlow are reshaping entire industries, helping the world to “run like Google.” As such, it’s perhaps not surprising that the company would announce today that it has assembled a who’s who of open source vendors, including MongoDB, Confluent, and DataStax, to run its software as wholly managed services on Google Cloud Platform (GCP). It would be convenient to think of this as some marketing master stroke to position “hero” GCP against “villain” AWS. It would also be wrong.
As Google head of open source Chris DiBona told me in an interview, such partnerships don’t represent “some sort of generous magical deal,” but rather simply a matter of “giving customers what they want.”
SEE: Google Cloud Platform: An insider’s guide (free PDF) (TechRepublic)
Not an AWS thing
Open source vendors have spent the last year casting AWS as the anti-Robin Hood, stealing from poor open source projects (them) to give to the rich (AWS). The accusation? That AWS “strip mines” open source projects, building managed services around them without contributing anything back. Never mind that no one has been able to point to a single instance of harm from AWS’ introduction of a new open source service, or that these plaintiffs have been booming even as they claim AWS will bust their businesses, and never mind that AWS has apparently tried to contribute to some of the projects, but without success.
Never mind all of that. It’s a convenient fiction and an easy marketing message, one that Google couldn’t completely avoid, touting the partnerships as evidence of a “much different approach to open source than our competitors.” Further, Google noted, “We’ve always seen our friends in the open-source community as equal collaborators, and not simply a resource to be mined.”
This is the absolute least interesting reason to like this deal.
Of much greater interest is the fact that Google is giving existing and prospective customers an all-in-one platform to get all their favorite data management and analytics vendors (Confluent, MongoDB, Elastic.co, Neo4j, Redis Labs, InfluxData, and Datastax) in a single place, coupled with existing Google offerings like BigQuery. While some of these vendors (like MongoDB) already have excellent managed services, the prospect of being able to process and analyze enterprise data in one place, while centralizing the user interface, billing, and support, is tasty, indeed.
For those looking for anti-AWS conspiracy theories, DiBona laid them to rest: “This is a business thing. We have customers, and they have money. If you just see this as about GCP and giving customers what they want, then it’s easy to understand.”
What customers want
That’s an interesting thought, “giving customers what they want,” as every cloud vendor claims to do that. In fact, no one waves “customer requests” around like a banner more than AWS. In this thing, however, Google may be onto something.
For example, it’s clear that enterprises want MongoDB–for over a decade, it has continued to rise in popularity. AWS initially tried to meet this demand with its own document database service, just as Microsoft did with Cosmos. AWS eventually settled on DocumentDB, an API-compatible database service that seeks to emulate (older versions of) MongoDB. Why? Because…customers.
SEE: Open source vs. proprietary software: A look at the pros and cons (Tech Pro Research)
As AWS vice president Shawn Bice said at the time, DocumentDB was “designed from the ground up to give customers the performance, scalability, and availability they need when operating mission-critical MongoDB workloads at scale.” What it wasn’t designed to do was to work natively with MongoDB’s latest code, in concert with the database vendor.
Google went a step further, figuring that if customers want MongoDB, they probably want the full-fat version, which includes both the code and the engineering support behind it. So that’s what they’re delivering, with the willing participation of MongoDB. It’s the sort of deal that both AWS and Microsoft could do, but neither has. Google, which has been criticized for years as not being enterprise-y enough, and not attuned enough to real-world enterprise demands, has reversed the narrative with this move.
It has also further demonstrated that no one feeds enterprise demand for open source better than Google does.
Open sourcing Google
Again, this isn’t because Google is some generous benefactor, doling out cash to open source companies so that they can keep the lights on. As DiBona stressed, “We’re showing that we know how to work with people to give our customers the best infrastructure possible for their applications.” What was left unsaid is that increasingly enterprises demand that their infrastructure and their applications be open source. By pairing its impressive infrastructure with these open source vendors, Google is, once again, demonstrating that it understands the value of open source on-ramps.
For example, DiBona went on, “If you’re using MongoDB but not yet GCP, GCP now looks more appealing. You can then try some of the other GCP services.” Come for the MongoDB service, stay for Compute Engine.
SEE: Why it’s pointless to criticize Amazon for being “bad” at open source (TechRepublic)
It’s hard to know exactly when Google recognized the value of open source as a way to stoke demand for its cloud infrastructure, but today the link is easier to see. Kubernetes, for example, started out with zero customers, but now more than half of Fortune 100 enterprises use it, according to Redmonk data. In July 2018, the Kubernetes container registry host served 129,537,369 container image downloads of core Kubernetes components, or greater than four million per day. This largesse is fueled by 300 different companies and over 2,500 committers (with a whopping 28,000 contributors). Google remains the largest corporate contributor, but it no longer has to shoulder all of the burden.
What it does get to do is profit from this community momentum. Today the vast majority (over 80%) of Google Compute Engine customers are also using Google Kubernetes Engine. Yes, AWS and Microsoft also profit from Kubernetes, but that’s not really the point. As evidenced in this partnership play, Google doesn’t have a scarcity mindset: There is lots (and lots) of money to be made in the cloud, and more will be made by sharing the wealth than by hoarding it.
And so we have Google, not interested in a “generous magical deal,” delivering that magic all the same by contributing heavily to open source, both in terms of code and now in terms of cash. It’s a great sign that Google Cloud Platform offers robust reasons for enterprises to look to it when considering how and where to run their open source software, making its claim credible to be the “best destination for open source-based services.”