Database decisions: AWS has changed the game for IT

You may not have heard of OpenSCG, but Amazon Web Services has. A week ago, AWS quietly acquired the PostgreSQL migration services company founded by PostgreSQL veteran Denis Lussier. While some PostgreSQL fans weren’t happy about the move, the OpenSCG acquisition is emblematic of a much larger move by AWS to serve a wide array of database needs.

At the recent AWS Summit, Amazon CTO Werner Vogels said as much, declaring that “what makes AWS unique is the data we have, and the quality of that data.” Taking a slap at Oracle in particular, Vogels derided the “so-called database company” for offering far fewer relational database services than AWS, and just a fraction of the array of database services that AWS offers (including NoSQL offerings).

With more than 64,000 databases migrated to AWS in just the last two years, AWS looks set to hold even more enterprise data.

AWS doesn’t tend to announce its acquisitions. They’re invariably small, not triggering any legal requirements to announce them, and while some companies acquire so they haveproducts to sell, AWS only acquires complements to the services it builds in-house.

Nor is it surprising that AWS would be interested in the PostgreSQL sponsor. As one Reddit commenter mentions, “True PostgreSQL expertise is difficult to come by and OpenSCG has a lot of it. If you combine that with Amazon’s clear support of deploying Postgres-related products (RDS/Aurora/Redshift) and its message of #DatabaseFreedom, … it becomes pretty clear why AWS was interested in OpenSCG.” Although OpenSCG has been an AWS partnerfor some time, OpenSCG has particular expertise in helping companies migrate to PostgreSQL.

Which is, of course, perfect for an AWS that is intent on moving orders of magnitude more database workloads than the current 64,000 to AWS.

AWS seeks to be the “every database” store

Not all those database workloads involve PostgreSQL, of course. Although the open source database has experienced a renaissance of popularityover the last few years, it’s just one of the various databases that AWS supports. AWS has been aggressively decomposing applications and infrastructure to give its customers the specialized services that let them develop what they want, Vogels says, “instead of AWS telling them what they must develop.”

You want PostgreSQL? AWS can help with that. How about a NoSQL database with infinite scale and predictable performance? AWS has that, too, with DynamoDB, but also through partners like MongoDB that run a large percentage of their workloads on AWS.

The list goes on.

And on.

All of which leads to the question “What does this mean for IT’s database decisions?”

The database choices aren’t like they used to be

Oracle and Microsoft’s trump cards to date have been that they collectively own three of the world’s most popular databases, including Oracle, MySQL (owned by Oracle), and Microsoft SQL Server. As data has changed, however, these trump cards have lost some of their luster, serving as an almost unwelcome crutch at times. Oracle has missed the market transition to big data applications.

By contrast, Microsoft has not rested on its laurels, releasing a spate of database options, including CosmosDB. Although Microsoft Azure has fewer database alternatives than AWS, it’s a strong No. 2 to AWS’s leadership position. So far, developers have preferred AWS’s approach, which is to offer maximum database choice, fitting particular databases to specialized needs. Even so, Microsoft at least has a credible strategy.

Oracle, by contrast, has spent years ignoring or deriding the cloud, then basically fork-lifting its database to the cloud. A year ago, it made the silly move of trying to raise the price of running Oracle on AWS, hoping to get customers to defect from AWS and run those workloads on Oracle’s struggling cloud. It hasn’t worked.

Nor will Oracle have much hope if AWS continues to move more database services into its arsenal of server-less functions. As industry expert Simon Wardley posits, “As Amazon’s server-less ecosystem grows, the more metadata it can mine, the faster its rates of innovation, customer focus, and efficiency. Once it gets to around 2 percent of the market then it’s game over for all those not playing at scale.”

Microsoft (and Google) are sprinting to add database services, including serverless options. Oracle keeps muddling through a 1980s way of thinking about the database, and it’s going to cost the database hegemon its lofty market position.

Meanwhile, AWS keeps steadily building out the database services developers require for next-generation applications, all while improving its abilities to migrate existing workloads to AWS.