I spent the better part of last week either traveling to or participating in Oracle Analyst World, an annual meeting of analysts and influencers that the company briefs twice annually en mass (Oracle OpenWorld is the other opportunity). Throughout the year, Oracle, and most other vendors, provides spot briefings on announcements and the like. This meeting offered a fire hose full of information about the company’s complete infrastructure stack and a bit about business applications.
At a gut level, I didn’t quite understand Oracle’s insistence on its engineered systems strategy. Conceptually, it makes some sense and it will probably make more sense in the future. The idea of hardware and software engineered together to produce a 1+1=3 situation is alluring but going into the meeting it had two obvious issues.
First, I was skeptical that enough time has passed since Oracle bought Sun to make the claim credible. There’s a lot of complexity involved in hardware, operating systems, databases, middleware and applications also, IT is mediated by standards so that multiple vendors can compete in the same space. However, going in I agreed there is money to be saved on the TCO front by reducing the time and effort required to integrate systems on a customer site.
But second, and perhaps most important, the universe of companies that really need solutions built together (let’s call this the “way better” solution for reasons that will become apparent momentarily) is much smaller than the universe of companies that need commodity systems mediated by standards (call this the “good enough” market). This sets up a two peaked bell curve in which there’s a luxury peak (for lack of a better term) and a commodity market peak.
If I wanted to violate my two issues stance, I would offer a third, which is that despite the rhetoric about systems engineered to work together, the company spends a lot of time assuring the world that its products can and do run on many configurations of hardware and middleware from other vendors. This is only appropriate given that Oracle operates in a heterogeneous world.
But even with all this in mind, Oracle executives brought out several proof points that are hard to dispute.
Point one — far better performance from hardware.
Point two — far better performance in database and analytics.
Point three — software becoming embedded in silicon for far, far better performance.
That said, it’s still a standards based, heterogeneous world and many sophisticated customers are quite happy to roll their own when it comes to their data center needs. A case in point is Salesforce.com, possibly Oracle’s largest customer. Salesforce CEO, Marc Benioff, is happy to tell you that his configuration is made up of commodity servers running some open source software as well as the Oracle database. Salesforce’s secret sauce is its own middleware that makes its systems so highly available and multi-tenant.
But the new solutions that Oracle is rolling out, especially in hardware and the middle of the stack suggest that those days are ending — at least for huge enterprises that need to squeeze every drop of performance out of their systems and supply rigorous security and unfailing up time.
These customers are the ones that need way better performance and can afford to spend a million bucks or more on a compute or data server or an analytics box. That’s the luxury peak alluded to earlier. Commodity used to be a bad word but the quality and sophistication of the good enough peak is so compelling these days that it’s quite good for most needs and Salesforce is only one example.
On the other hand, you have to like and respect the Oracle’s scrappy pursuit of business especially in older traditional markets like database and hardware.
This is important because Oracle isn’t simply offering a few percentage points of improvement, they are truly setting a new paradigm by providing hundreds of percent performance increases and that moves the needle for markets and customers. I lost count of the times we were told in the briefings that a new device or version of software drives two to seven times better performance. When that happens, you aren’t improving within the paradigm, you are inventing a new paradigm and a new performance curve. This is curve jumping. And when that happens you are hitting the reset button on the entire market.
For Oracle, this means a new 24 second clock, a new set of downs, extra innings. Whatever the metaphor, the practical effect is the same and all this comes along at an opportune time. The Big Data and Cloud Computing revolutions have made it essential for vendors to offer new performance levels. But given all this, I think there is now an ethical concern that we all have to deal with, a concern that has never existed in any of the other paradigm shifts or curve jumps that we have yet seen in IT. That’s the subject of my next post.
(Cross-posted @ Beagle Research, LLC)