The cloud was supposed to kill open source. Instead, savvy cloud operators appear to be using open source as an on-ramp to proprietary services, giving them reason to increase investments in complementary open source projects. Google is the obvious example, spinning out TensorFlow and Kubernetes as a way to raise a generation of developers anxious to perfect machine learning and container-driven workloads on the Google Cloud Platform.
But Google isn’t alone. It turns out that Amazon Web Services has its own open source strategy, one perhaps less obvious but no less potent. From my conversations with AWS customer InfoScout, AWS seems to be building natural bridges between on-premises databases like MySQL and cloud services like Amazon Aurora, giving customers a reason to start with open source but finish with AWS.
Moving from MySQL on prem to Aurora in the cloud
Just a few years back, one prominent criticism of public cloud IaaS platforms like AWS was that while enterprises might be happy to run dev-and-test workloads in the cloud, production would always happen within private datacenters.
That might have been true for a nanosecond, but it’s demonstrably, outrageously false today—companies like Capital One are dumping most of their datacenters to push the majority of workloads to the cloud. Concerns over governance and risk have given way to public clouds like AWS based on their security model, ability to provision infrastructure on the fly, elasticity to handle purchasing demands at peak times, high availability, and the pace of innovation.
Oh, how the times have changed.
Indeed, changed so much that now it is the public cloud that captures production workloads while dev-and-test workloads run on-premises on developer laptops. Nowhere is this shift more glaring than in databases, which have been the most resistant to moving to the cloud, given data gravity.
As InfoScout’s senior director of engineering, Dana Ford, told me, while the company embraced MySQL because “it has been around for decades,” a proven database for a variety of workloads. Scaling it in production, however, stretched the limits of the company’s database expertise. InfoScout couldn’t afford for its staff to become expert DBAs.
So it went with Amazon Aurora.
“Aurora speaks the same MySQL dialect [offering wire-to-wire protocol compatibility],” InfoScout CTO John Brelig explained, “This allows us to still have dev environments with MySQL but then deploy production to Aurora almost seamlessly.” When pressed on how almost that “almost” qualifier is, he responded that it’s more a matter of Aurora not being in lockstep with MySQL versions (Aurora doesn’t currently support MySQL 8.0 and only recently started to support Version 5.7).
But it also comes down to how a company defines the database and the schema structure of its tables. In InfoScout’s case, the transitioning MySQL development workloads to Aurora production workloads requires “some tweaks.” Nothing major.
Letting enterprises invest in customers, not DBAs
Cue the diabolical laughter of the evil Jassy overlord, bent on luring unsuspecting developers into his proprietary lair. I’m sure some will see this MySQL-to-Aurora play as just that—a sinister plan—but hearing InfoScout’s engineering executives talk about it, they simply don’t see it that way. As Ford said: “InfoScout values the cloud and Aurora so much because it allows us to focus on our business and less on the code and infrastructure.”
I heard this same theme over and over as we spoke. While both Brelig and Ford were quick to highlight how much InfoScout depends on open source, they were equally quick to talk up the value they derive from building on AWS: “We love AWS and Aurora because almost at the flip of a switch we were able to focus more on customers. … We’d default to open source but we also look at other solutions that give us more time to do what we do best for our customers.”
It’s the business agility that derives from the cloud that enables companies like InfoScout to spend more time creating business logic to help with customer needs, rather than becoming master DBAs. Open source is great for enabling parity between development and production environments. But cloud services like Aurora make that partnership even better by removing much of the heavy lifting associated with scaling an open source database like MySQL in production.
It’s the best of both worlds. Yes, it also leads to lockin to a particular cloud’s database services. But that’s a trade-off a rising percentage of enterprises seem happy to make.