Read the stories about the IT world and you can’t escape a couple of truths. One is that hardware is commoditizing; IBM sold its x86 business and everyone is talking commodity servers. Another truth a bit harder to pick out but no less real is that platforms in the broader sense are definitely coming into their own.
A lot of this is a sort of industrial inevitability. If you make auto parts by stamping them out in the millions you don’t create a do-it-yourself auto industry (“See that bin over thar? Do your thing!”), you simply change the limiting factors. Buyers who see their server costs fall by half are dismayed to find out their IT costs didn’t fall much, but that’s because other stuff that added up to more was still there in the picture.
Platforms are my name for the plug-and-play mission-based configurations that could be hardware/software or just software designed for commodity hardware. The idea is that somebody with a problem can buy a prefab solution. Since somebody with a problem is likely somebody with a budget, that adds up to sales success. And the more complicated the path from basic tools to solutions is, the more valuable the platform is.
For the cloud, SDN, NFV, and even things like our “Internet of Things” we’re obviously complicated enough for all practical purposes. That’s what I think is behind the recent VMware/EMC strategy to create platforms. Not only do they ease the problem-to-solution transition they also guarantee a vendor with a lot of piece parts an opportunity to easily fight off best-of-breed-driven fragmentation of the opportunity.
VMware isn’t the only company who is proposing platform-based selling. Red Hat’s Enterprise Linux has been a platform from the first. Recently, though, others seem to be taking platformization to a more detailed level. VMware is one, but another example is the Intel/Wind River Carrier Grade Communications Server. Cloud platforms, carrier platforms—both are examples of a refinement to a general platform model. It’s not going to stop there either; IBM, HP, and Oracle are all working to an almost vertical-refined platform strategy, because it works.
The interesting question this all raises is how pervasive the trend will get, and how that will impact the vendors. In particular, you have to wonder how cloud, SDN, and NFV strategies will be impacted by platformization, either in a positive sense (somebody is promoted by the trend) or a negative one (somebody is marginalized).
A data center solves an IT problem, not a computer or software. If platforms are good then it would seem that the trend could benefit players like HP, IBM, Oracle, and SAP who could offer a fairly complete strategy if they adopted open-source tools as a foundation. The challenge is that most everyone has data centers already, so you have to look not so much at what solves IT problems as what solves new IT problems—ones that don’t already have some data center support in place. That’s why things like the cloud, SDN, and NFV are important. They are incremental issues that admit to incremental solutions, and that opens the door for packages.
The obvious question is “What kind of packages?” I agree that “cloud” is one of those incremental issues, but singing “Cloud!” at the top your lungs doesn’t make you an opera star. The key to “package success” is to get astride a compelling value proposition. Where are they to be found? It’s never easy, but in some areas it’s easier than others.
SDN and NFV, according to buyers, are justified on cost management and agility. In the main, those are operational issues, which suggests that the sweet spot in package deployment for SDN and NFV lies on the operations side. At present, nobody has made a compelling case for having addressed that space, as I’ve noted in prior blogs. We’re stuck in “structural orchestration” and we need to rise up to the functional level to augment what we have.
In the cloud, it’s more complicated. I think you can make a case for functional-based operations changes there, but the question is whether a cloud based only on cost reduction can ever be really successful. Benefit augmentation is always safer, but at least for now we have a notable lack of insight into what a benefit-driven cloud might look like, or how it might be different from what we have.
A cloud driven by benefits is a cloud that does new stuff, things we either can’t do effectively today or can’t do at all. We’ve had virtually no dialog on what this kind of new stuff might look like. It seems likely that the benefit cloud would be a kind of PaaS on which valuable applications and services would be created, but without knowing anything about the specific applications and services, it’s hard to say what the specific features of the PaaS might be.
It’s also hard to say whether this “benefit cloud” might not nibble at the edges of the portion of the SDN and NFV value propositions that relate to “Service Agility.” Does a platform that facilitates value-based applications not also provide support for new service revenues? The cloud, in its public cloud instantiation, is all about service revenues.
The people who have to care about this sort of thing are on either the leading edge or the trailing edge of the platform revolution. Red Hat, Wind River and VMware are arguably the best-positioned to create a benefit-driven platform, no matter whether you call the effort “SDN”, “NFV”, or “the cloud”. Alcatel-Lucent, Cisco, HP, and IBM are all somewhat at risk, but for different reasons. Alcatel-Lucent and Cisco have specific software issues that make it harder for them to present themselves as a platform vendor. Who thinks of either as software powerhouses? Alcatel-Lucent, at least, has some obvious platform directions out of mobile services and even NFV. Cisco can say “Cloud” and they’re one of many.
HP and IBM have too much legacy in the IT and applications area, and are inclined to bee to conservative with respect to platform positioning. All these guys, though, have the common problem of “toppyness”; there are too many different application- or service-linked choices to make or defend. It’s easier if you have some middle-level option (like operationalization), which is why I think Cisco needs to retune its cloud, SDN, and NFV story to grab what they can there. Apart from the risk that VMware obviously continues to generate.
I can’t say whether platforms are for-sure the way of the future. I can say that they’re a way that somebody can control the future. Let’s see who, if anyone, steps up!