Is Jassy just jazzing on AWS database migration numbers? Smells fishy...
We're not saying Larry Ellison has a point, but...
Analysis Amazon Web Services chief Andy Jassy doesn't often tweet, but when he does it's usually to applaud the popularity of AWS's database migration service.
On New Year's Eve Jassy bragged that the service had captured more than 16,000 databases from its rivals since launch on 15 March, 2016. According to him, earlier this month, the DMS has completed more than 20,000. Just this week he claimed 22,000 migrations of either Oracle, SQL Server, MySQL, MariaDB, or PostgreSQL databases "with virtually no downtime".
AWS exceeded its $10bn annual revenue target in its last year and Amazon reported over $3bn in operating income for its web services segment. Unfortunately, the corporation doesn't publish granular data on its revenue generation and so it's difficult to know in precise fiscal terms how much of that cash came from cloud databases or those database migrations.
All we have to go on is the Jassy tweet, which is absolutely unverifiable. Even assuming Jassy's statements are true, questions still remain about the size and importance of the databases that have been migrated to AWS's cloud, and this to Redshift or AWS's Relational Database Service.
It's clear who AWS is going after. The same RDBMS giant all challengers have chased over the years: Oracle, owner of half that market. That much is is evident from Jassy's hashtag #DBfreedom.
#DBfreedom is a clear reference to the clingy relationship many Oracle customers have found themselves in when seeking to escape their expensive hardware for the less onerously scalable and manageable morass of the cloud. Thanks to the sticky nature of data, who you chose as your database provider is a strategic platform choice, not your standard product decision. But for many, thanks to more than a decade of acquisitions in apps and hardware, Oracle has tightened its grip on its customers' infrastructure – and their budgets. No wonder Oracle customers are perceived as willing to move if not actually jumping.
But the degree to which AWS customers are truly free on AWS is hotly contested.
Oracle executive chairman Larry Ellison stirred the coals recently when he said Amazon's database products and apps built on AWS stay on AWS: "Build an app on Redshift and you will be running it forever on Amazon – you are locked in, baby," he said. "So if Amazon raises its prices you better get out your chequebook."
James Hall, director at AWS evangelist customer Parallax, rejected Ellison's quote as "hyperbole for a few reasons". "Firstly, Redshift is largely Postgres-compatible," said Hall. He acknowledged, though, that "if you really need to switch for some reason, it's still an engineering effort."
Hall also rejected Ellison's scaremongering on prices. AWS maintains that it cut its prices 53 times since 2006. "If Amazon did hike their prices... you could hop along to another hosting provider, taking advantage of the schema migration."
But this opinion is not shared by everybody and the mounting problem for AWS customers is not list prices that have dropped, but rather the total amount they end up shelling on AWS as a consequence of using so many different AWS services, accidentally leaving instances running, and the simple act of creating, housing and accessing ever greater volumes of data on AWS.
Conscious of this, AWS last year announced a budget calculator to help customers control spend, setting up alerts as they approach their budget's limit.
And last year AWS felt compelled to deliver a presentation at its annual Re:Invent conference entitled Busting the myth of vendor lock-in.
The UK's Competition Markets Authority last month wrangled a commitment from Amazon along with Apple and Microsoft to provide cloud storage users with fairer contracts.
That followed a CMA review in response to consumers being "surprised by significant price increases and by reductions to unlimited storage capacity deals after contracts have been agreed". Such activities, as per Ellison's incitement to apprehension regarding AWS, would face significant regulatory attention.
That said, cloudependance has become more of an issue for analysts and media than for customers who seem sanguine about commitment and lock-in.
Indeed, customers The Reg has talked to are developing sophisticated dual-cloud strategies. That involves shopping around using, say, AWS for compute and Google for database, rather than sinking everything into a single provider as happened in enterprise software's old, on-prem days.
Of course, AWS's entire business model is built around keeping costs low, and it potentially does make sense for some customers to consolidate their entire operational estate within the confines of the cloudy cage, and as much as it pains us to say it, this does not invalidate Larry Ellison's argument.
If and when AWS does raise prices, customers who have committed too much to its cloud are going to have to take stock. Decide whether they can absorb the increase and measure the value they receive from AWS and against that assess whether Hall's "engineering effort" would cost them more than the savings they might potentially make. ®