Oracle’s biggest cloud customer highlights its biggest cloud problem

Only seven years after Amazon Web Services (s amzn) debuted and the Software-as-a-Service business model blew up, Oracle has finally come around on cloud computing with its flagship database product. On Monday, it revealed its anticipated Oracle Database 12c edition, calling it “the first database designed for the cloud.” Personally, I think that depends how you define cloud.
Oracle 12c does offer up a slew of features, some of which — such as multitenancy and geographically distributed load balancing/failover — are very cloudy. If I were building a cloud service or looking to update an existing one, I’m sure these would appeal to me. Heck, I might even demand them.
oracle mt
That being said, if I’m building a Software-as-a-Service startup that’s doing nothing now but that I hope will spread like wildfire, the $47,500 per processor list price of Oracle 12c Enterprise Edition — the only edition offering multitenancy (for an additional fee) — might be a little steep. And even if I were willing and able to pay that price, it also means I’d have to deal with physical servers somewhere or, if the reports are true, host my database on Microsoft’s Windows Azure cloud. And if I needed to make changes to my database layer to account for — or enable — changes to the application, well, I might want something a little more flexible.

A tale of two architectures

To me, at least, the kind of company I just described is the kind of company that exemplifies cloud computing. It’s about scrappy startups building applications on AWS, growing like crazy, and engineering themselves out of the technological hole that success brought along with it. The essence of cloud computing, in this sense, is freedom — freedom to pick the tools that are best, to tinker with them until they fit your needs, and to swap them out entirely if they don’t work or if they’re suddenly too lightweight for the job (a reason why Feedly just replaced MySQL with HBase).
In this version of the cloud, open source reigns supreme, and not just for small companies. At our Structure conference in June, Netflix cloud architect Adrian Cockroft told me the company has all but weaned itself from enterprise software to power its streaming business, and it’s certainly not looking to bring on any more. If Facebook, Google, Twitter and the rest of the new web power players can’t find what they need in the form of open source software, they just build it themselves (and then subsequently open source it).

Netflix's data architecture.

Netflix’s data architecture.

To be fair, though, the kind of customers Oracle is expecting to attract with 12c probably don’t much care about the things that startups and web powerhouses do. Global companies with big, important applications need big, important databases. If Oracle is the most secure, most sturdy and highest-performance option around — or if it’s just too much work to switch — it’s arguably worth the money., for example, can standardize on expensive Oracle software and hardware because its user growth has become predictable (but is impressive nonetheless) and it’s probably a little late to switch from the database it was built on without some Grade A engineers on board. That would be an expensive and timely proposition, especially because any customer downtime would result in public outcry.
Heroku, the 2010 Platform-as-a-Service acquisition that was founded eight years after Heroku and now runs more than 3 million application? It’s built on PostgreSQL and it’s constantly updating its features as that technology matures.

No one ever got fired for buying Oracle. But did anyone ever get promoted for it?

There’s cloud in the sense that the architecture is multitenant, distributed and, ideally, that the application is delivered as a service. So what if it runs on Exalogic boxes or you can’t tweak the code or that it’s going to cost a boatload to run over its lifetime?
And then, there’s cloud in the sense that it’s all of that first set of characteristics, but none of the latter. The cloud where a team of smart software engineers and architects can turn an open source database — or two or three — and some generic boxes into the foundation of a product that can evolve and scale to its heart’s content without costing an arm and a leg in upfront costs.
Neither approach is fundamentally right or wrong, but choosing one over the other does suggest how a company views technology as a tool for innovation, as well as the stage of its business. There’s planning for the future and there’s planning to maintain, maybe ratchet up, the status quo. An awful lot of seemingly smart companies are choosing the former.