With this week's announcement of Oracle Cloud, the vendor has finally reached the halfway stage in the journey that former company president Charles Phillips outlined to me at Oracle OpenWorld five years ago. Oracle was helping its customers take "stair-steps to on-demand," Phillips told me then. Five years later, Oracle has finally managed to climb halfway up the ladder. Let's hope the rest of the ascent will be faster.
Unfortunately, today Oracle remains stuck at the middle rung with its stubborn insistence on dedicating individual database instances — and sometimes entire application stacks — to individual customers. These dilutions of the multi-tenant ideal serve the database vendor's self-interest, because it gets to sell more licenses, but in cloud terms it's half-baked. Interestingly, the arguments in favor of this approach have not advanced since Phillips' conversation with me back in 2007. He even used the same word, 'co-mingling', that Ellison used this week to imply that full-stack multi-tenancy is somehow less safe than confining customers to separate databases or servers.
In one of my frequent displays of inadvertent clairvoyance, this is a line of argument I already dealt with last week, but since it's been trotted out again, I'll reiterate what I said then:
"Do you worry about your data co-mingling with others when the packets pass through the network's routers? You know that the headers on the packets make sure that your data won't accidentally go to someone else's endpoint. Cloud vendors use exactly the same logical separation to keep your data from 'co-mingling' with anyone else's. The fact that it may be stored on the same disk or go through the same processor chip is as irrelevant as worrying about sending your physical mail through the same postal system as your competitors."
Recommending single tenancy as a way of keeping your data safe from 'co-mingling' in a cloud environment is akin to recommending abstinence as a means of avoiding teenage pregnancies. It's somewhat old-fashioned given the better targeted methods of protection available today; but worse than that it's doomed to failure because it ignores the realities of how the world works. In a cloud environment (even Oracle Cloud), co-mingling is the norm and your data is going to run through the same cables, routers and management servers as everyone else's even if it resides on separate databases. Your employee's logins are going to be 'co-mingling' with those of partners, customers and suppliers that they collaborate with as part of their day-to-day projects. You're going to need a more sophisticated form of security than merely cloistering your data in a dedicated instance.
But at least Oracle has been consistent in its messaging, not merely five years back but all the way back to 1999, when it launched its Business OnLine application hosting operation (later renamed Oracle OnDemand). My report of the New York launch is no longer online, but according to my archive copy, Ellison then predicted, "The software business is on its way to becoming a service business ... If you don't understand this [as a software vendor], you're going to be in a lot of trouble."
Given that background and my conversation with Phillips five years ago, I'm more inclined than some of my analyst colleagues to give credence to Ellison's claim this week that Oracle Cloud is the result of a seven-year development project. Back in 1999, Jonathan Lee, the founding CEO of Corio, one of the earliest ASPs or application service providers, was warning that vendors like Oracle would have to fundamentally rearchitect their products to survive the shift online. Project Fusion had many objectives, of course, and presenting it as primarily focused on a cloud future is a substantial reslanting of history. But one of its benefits was the implementation of a service-oriented application platform that is capable of operating in a cloud environment far more efficiently that what went before. Indeed, looking at the architecture slide of Oracle Cloud presented this week, the main difference I can see from slides of the Business OnLine and Oracle OnDemand architectures ten years back is that the applications have become services — a prerequisite for the transformation Lee had been calling for.
The problem for Oracle Cloud is that the majority of the vendor's customers still want to host their applications on-premise and therefore the Fusion applications can never be tuned to operate efficiently as massively scalable multi-tenant instances. As pureply cloud providers gain the upper hand, that will put the product set at a massive competitive disadvantage. But that's for the future. For now, in view of the far higher license revenues that Oracle can achieve so long as its customers are confined to single-tenant instances, there is no way Oracle's president and CFO Safra Catz would allow Ellison to talk up multi-tenancy even if he wanted to. So instead he'll continue to trot out the co-mingling claptrap and all the other received wisdoms that enterprise IT buyers continue to lap up because neither do they want the disruption that a full embrace of cloud would cause their own empires.
As a majority owner of NetSuite and a significant minor investor in Salesforce.com, Ellison has hedged his personal bets on cloud in any case. If Oracle can remain plump on the ingrained buying habits of dyed-in-the-wool CIOs by continuing to sell them on-premise implementations and single-tenancy cloud, he's merely discharging his fiduciary duty as Oracle's CEO in encouraging them to do so. Why should he care? Perhaps in his heart, though, he's irritated at the shortcomings of Oracle's middle-rung cloud and that's why his claims of Oracle superiority are starting to lose their edge.