Don't fear vendor lock-in, Oracle says.

Of course, any company of Oracle's scope would say that. But the complexity of the cloud doesn't lend itself to a DIY approach, said Dave Donatelli, executive vice president of converged infrastructure, during his Oracle OpenWorld keynote on Wednesday.

Donatelli was talking about hyperconverged infrastructure, but Oracle's idea of "converged" extends further than some companies'. In most discussions, "you don't really get above the network layer," he said.

Oracle, of course, offers plenty that's above the network layer, in the form of databases and applications and the middleware to connect it all.

Hence, Oracle's pitch is that it's better to buy all of this from one vendor than to assemble best-of-breed piece-parts. It's a familiar argument. Many a hardware vendor has pointed out that Google popularized DIY cloud infrastructure — designing its own servers and switches, and applying SDN to it all — needed an army of R&D troops to do it.

Donatelli reiterated those points and gave examples of DIY failure. In one, a $30 billion-a-year financial services company just couldn't get a DIY cloud project to work. Scrapping that idea, the company decided to run its Oracle software (databases, PeopleSoft, and the e-business suite) on Oracle hardware. The result was an operational savings of $42 million, Donatelli claimed.

Another alternative he pooh-poohed was the use of Amazon Web Services (AWS) mixed with a company's own private cloud. Those are two different environments, of course, and call for different tools and processes. "The burden falls on you. You have to make this all work," Donatelli said.

Oracle, by contrast, could be the foundation for a private and public cloud, supplying the same architecture on both sides, he said. (Note that Linux containers are designed to get around this problem by making it trivial to move an entire workload from one environment to another.)

The correctness of the vertical integration argument will always be debatable. But the idea is timely on a couple of fronts.

First, the panoply of cloud options opens an opportunity for some company to act as a guide. Oracle's argument isn't much different from Cisco's assertion that customers now prefer architectures to point products.

Something similar is happening in SDN and NFV. An OpenWorld panel about NFV's good, bad, and ugly facets pointed out that NFV lacks an end-to-end distribution. That's understandable, because NFV is so young, but you have to wonder if some kind of NFV "distribution" might emerge — something similar to an OpenStack distribution, where one company assembles all the pieces for you.

Second, "big" is in. Analysts expect cloud services, especially public clouds, to boil down to a handful of major players. And the tech industry's pendulum is swinging toward consolidation, with pending megadeals such as Nokia buying Alcatel-Lucent, Avago acquiring Broadcom, and Dell acquiring EMC.

Big companies love vertical integration. It seems likely we'll hear a lot more about integration, or architectures, or one-stop shops.