Data Centre


Hubris, thy name is Oracle: So, cloud is still totally for nerds, right?

12 new data centres is too little, too late

By Matt Asay


Comment If Larry Ellison earned a dollar for every cloud-hyped phrase he made, Oracle's market share in cloud infrastructure wouldn't be the miserly 0.3 per cent – by Gartner's calculations – that it is today.

Big Red has spent years playing make-believe with cloud, trying to convince the world that it could simultaneously underinvest in data centres while reaping maximum profit. Oracle is finally learning that it must invest in order to collect on its cloud ambitions.

As such, the company recently announced that it will build 12 new data centres, despite claiming it didn't need more because its databases and computers are so much faster than those at AWS or Microsoft Azure. Even with this loosening of pursestrings, it's unclear whether Oracle has much chance of bridging the gap with its cloudier competitors.

Pot, meet kettle

Oracle has become more caricature than competitor, regularly spouting off in ways that makes its executive suite seem foolish at best. Back in 2009, as an excuse to start falling behind Amazon Web Services, then-CEO Ellison ranted: "Everyone looks around and is like, 'Yah! Like everything is in the cloud.' My objection is it's absurdity – it's nonsense... What are you talking about? It's not water vapor. It's a computer attached to a network!"

And, of course, no one was better than Oracle at computers attached to networks.

The problem is that, in fact, "cloud" is neither "water vapor" nor as simple as attaching a few computers to a network. Oracle's unwillingness to take the trend seriously (along with peers at IBM and elsewhere) led AWS CEO Andy Jassy to say in 2017: "I don't think in our wildest dreams we thought we would have a six to seven-year head start." But they did, and Ellison's hubris played a big part in ensuring that.

Even as Oracle stiff-armed the future, its core licensing business was in terminal decline. As highlighted by Redmonk analyst Stephen O'Grady in 2013, Oracle has increasingly struggled to sell software to anyone except those already buying. Even its cloud strategy has targeted a relatively small niche: existing customers looking to shed data centre applications and move them to the cloud.

Unsurprisingly, in its last quarter, Oracle's new licence revenue was stunted, even as its SaaS business blossomed 55 per cent to $1.1bn and its overall PaaS business (which includes IaaS) went up just 21 per cent to $396m. This cloud business is growing faster than the declines in its legacy business, but not nearly fast enough to keep pace with market leaders AWS, Microsoft and Google.

If you build it, will they come?

Clearly, Oracle's strategy hasn't worked. That strategy, in part, has inexplicably depended on the firm managing to save its way to cloud glory. Even as AWS, Microsoft and Google shovelled $41.6bn into data centre build-outs last year, up 33 per cent over the year previous, to bring their collective total to hundreds of data centres, Oracle was hard-pressed to announce just three data centres in 2017, spending just over $2bn.

Even then, Oracle wanted the world to think this was all by design. CEO Mark Hurd tried to explain away Oracle's paltry investments in data centres: "If I have two-times faster computers, I don't need as many data centers. If I can speed up the database, maybe I need one fourth as many data centers." The reason for Oracle's parsimonious approach was simply that it's the greatest at everything. Never mind that for all its bragging about having "custom-built" infrastructure, so, too, does everyone else. The difference, however, is that AWS, Microsoft and Google have dramatically more scale and experience with custom-building their hardware and software. Oracle is a neophyte.

Keeping up with the cloudy Joneses

Can Oracle catch up? That's hard to imagine. While the company can (and must) dramatically expand its data centre footprint to even be credible, it's hard to see how it can catch the market leaders. Putting aside raw capacity, even Microsoft Azure, a distant second place to AWS in market share, has struggled to add features and services at the pace of AWS. Oracle isn't even in the same universe as AWS when it comes to cloud services, but also can't get into the same zip code as Azure or Google Cloud. Again, the database giant is hoping to lure existing customers with offers like automation across "pretty much" all its services, but has yet to demonstrate that it can compete head-to-head with the cloud heavies.

But Oracle is not simply relying on sizzle to sell this product – it has been tuning the sales machine, hiring away the sales and engineering staff of more experienced rivals while massively increasing bonuses and changing the structures to target renewals. Licence experts report Oracle customers being offered cloud as a means to bring down the bill on any licence non-compliance issue.

Small wonder, then, that Deutsche Bank Securities Inc analyst Karl Keirstead has declared that it's already "game over" for Oracle in the IaaS market.

Nor is Oracle's hitherto impregnable database fortress looking quite so secure these days. Modern data looks a lot more like Amazon DynamoDB or Microsoft Azure CosmosDB than the neat ones and zeroes of relational databases of yesteryear. Oracle – unlike AWS, Microsoft and Google – has no experience building next-generation cloud applications at massive scale so is always going to struggle to build modern data infrastructure.

As such, Amazon has helped migrate 50,000 database instances to AWS, much of them from Oracle, as AWS CEO Andy Jassy has proclaimed. The pace of those migrations seems to be accelerating to roughly 5,000 per month. As Gartner analyst Merv Adrian told me in an interview, Oracle has dropped database market share every year for the past four years. While the company still has 40 per cent of the market, its absolute dominance is no longer assured.

Is there any positive news for Oracle? Sure. While Oracle is a rounding error in IaaS and can muster 2 per cent market share in PaaS, it can claim a more respectable fourth-place SaaS with 5.6 per cent market share. Ellison may boast that Oracle is "completely transforming the way all companies buy and use cloud by providing flexibility and choice", but the best the company can claim is that it remains relevant in SaaS. That's a big market, and a big deal.

Meanwhile, Oracle gets to experiment with the cloud market because it's bankrolled by a massive installed base of enterprises paying maintenance on all those old database instances. A recent Rimini Street survey indicated that as much as 74 per cent of those customers run the legacy database unsupported, yet pay all the same. That's high-margin revenue Oracle can use to fund its aspirations in cloud.

Twelve more data centres is one way to spend that cash. Will it be enough? No. But it just may give Oracle the ability to convince more of its customer base to give it a chance with their cloud-leaning applications. That will be worth a few billion a year, even if the competition is counting in tens of billions now (AWS) or in the not-so-distant future (Microsoft and Google). ®

Matt Asay is Head of Developer Ecosystem at Adobe.

Sign up to our NewsletterGet IT in your inbox daily


More from The Register

Oracle gets busy with Lazy FPU fix, adds more CPU Spectre-protectors

Oracle Linux and VM get their innoculations

Oracle Database 18: Now in downloadable Linux flavour

Oh, and Windows, but cool kids don't use that

Oracle: Run, don't walk, to patch this critical Database takeover bug

Flaw in House Larry's flagship product allows 'complete compromise' of servers

Detroit sh*t shifter's operating costs waste away with Oracle's cloud

Sewerage department pinches off big brown puff for Big Red

Firefighters choke on Oracle's alleged smoke-and-mirrors cloud

Pension fund cries fraud over database giant's boasts about its off-prem biz performance

Is someone chopping onions? Oracle cloud boss bids colleagues emotional farewell

Thomas Kurian to take 'extended leave' from Big Red

No do-overs! Appeals court won’t hear $8.8bn Oracle v Google rehash

Only thing left now is a Supreme Court bid in row over Android and Java copyright

Oracle puts release of new freebie mini-database on ice to work out kinks

Issues 'too severe' to launch this summer

Oracle tells students: You're not going to solve the world's problems – but AI and ML might

Big Red's free educational curriculum gets 2018 reboot

Thought Patch Tuesday was a load? You'll want to avoid this Oracle mega-advisory then

But you'll definitely want to check out the libssh bug