Amazon says its consumer division has dumped Oracle's database platform, and the company will offload most workloads by year's end. Will it inspire AWS customers to do the same?
Vendor braggadocio is a staple of the tech industry, and while best heard with a skeptical ear, sometimes it raises important questions for enterprise customers to consider.
At last month's OpenWorld conference, Oracle CTO Larry Ellison renewed his claim of Oracle's superiority over AWS, not only in cloud infrastructure but also cloud security, and in particular, cloud database technology. Oracle's database platform is entrenched in the enterprise computing landscape, though emerging competitors have chipped away at its market share.
Ellison boasted that Amazon remains an Oracle database customer, as does SAP for both cloud services and on-premises customers. But as is often the case with Ellison, he bent the truth a little. Although it's true that many SAP ERP implementations use Oracle as an underlying store, SAP has ported some applications to its own HANA in-memory database.
In response, AWS CEO Andy Jassy tweeted last week that Amazon's long-rumored cloud database migration off Oracle and onto its own data management products are well underway -- Amazon's consumer business turned off its Oracle data warehouse Nov. 1 and moved to Redshift.
In latest episode of "uh huh, keep talkin' Larry," Amazon’s Consumer business turned off its Oracle data warehouse Nov 1 and moved to Redshift. By end of 2018, they'll have 88% of their Oracle DBs (and 97% of critical system DBs) moved to Aurora and DynamoDB. #DBFreedom
Amazon's Oracle data warehouse was one of the largest (if not THE largest) in the world. RIP. We have moved on to newer, faster, more reliable, more agile, more versatile technology at more lower cost and higher scale. #AWS Redshift FTW! https://t.co/AE50r7MUmW
If accurately conveyed -- and there's no public reason to doubt Jassy and Vogels' claims -- Amazon's cloud database migration effort is towering, and its customers could benefit from the lessons it learned along the way, should the company choose to share them. The timing couldn't be better, either -- with the massive AWS re:Invent conference just weeks away, it wouldn't be a surprise to see Jassy and Vogels revisit this topic during their keynotes.
Beware of database migration unicorns
That's not to say any enterprise customer could replicate AWS' experience to migrate off a major platform vendor, given its vast engineering resources to throw at the task. Nor is AWS itself immune to major customers abandoning its ship.
And there are broader considerations. Beyond the performance of the underlying database service, applications and analytic data pipelines tie into the broader world of compute, storage, integration and developer services, and are deployed and managed globally.
"The world does not revolve around database services in isolation," said Doug Henschen, VP and principal analyst at Constellation Research, based in Cupertino, Calif. "When companies go cloud they look at the total ecosystem of services and available capacity."
Moreover, there's a good reason why the industry evolved around the Oracle stack. "Amazon was founded in 1994 and Amazon Web Services was launched in 2006, and Salesforce was founded in 1999, long before real alternatives existed," Henschen said. "Once you go down that path, it's not a choice you can easily unwind years later after building on that foundation."
Sometimes, it doesn't matter which [database] product you have as long as it's reasonably good.
Curt Monashpresident, Monash Research
The same goes for enterprise shops with a wealth of bespoke applications built and maintained with Oracle. If AWS uses re:Invent as a forum to discuss Amazon's cloud database migrations off Oracle and onto its own platform, customers in that situation will want to hear how the company can help them do it, too.
Customers that start from scratch must weigh other tradeoffs. PostgreSQL has gained ground for transactional workloads, with services available on all major public clouds, as well as support for on-premises deployments through EnterpriseDB, Henschen said. Amazon Aurora is compatible with MySQL workloads and PostgreSQL, but is only available on AWS.
Oracle database costs are famously substantial, so the company has pushed to differentiate through its new autonomous database capabilities, not just performance and scalability, to address competitive threats posed by cloud and open source rivals, Henschen said.
Beyond his claim of performance superiority over Amazon Redshift and Aurora, Ellison has also said Oracle's database is much cheaper to run on the company's IaaS than Amazon. Of course, customers should judge for themselves with benchmarks that reflect their real-life workloads.
Overall, there's a lot of FUD in the air between AWS and Oracle's clouds. And decisions on large-scale database migrations like the one Amazon has undertaken can't be made lightly.
"DBMSes have many use cases, and some are more sophisticated than others," said Curt Monash, president of Monash Research in Acton, Mass. "Sometimes, it doesn't matter which product you have as long as it's reasonably good."
Any database platform migration will require refactoring, more so depending on their sophistication and complexity, Monash said. "If you're not prepared to undergo that expense and effort then migration is dangerous."
Inventory and order management, which lie at the heart of Amazon's consumer business, are two of the most intense database use cases. Amazon pulled the trigger on its major cloud database migration not only because it's good PR for its own services, and bad PR for Oracle, but because its cost-benefit analysis pointed that way. At re:Invent, we may see how the notion resonates with AWS' own customers.
