Whatever you want to call it, we’re having a revolution in IT. HP’s announcement of a cloud-friendly architecture to build data centers, one that includes prefabricated provisioning tools, was mirrored in many ways by the IBM PureSystems launch yesterday. I think IBM is right in that the new wave is the most important thing in IT in decades, but I also think IBM could have done a better job of explaining not only the “why?” but even just what the heck they actually did.
Underneath all of this stuff from HP and IBM is an interesting truth, which is that it’s getting more costly to support IT than it is to buy the gear. This is particularly true for SMBs who lack the large IT organizations to create career paths for IT professionals, and who often have only occasional need for key specialties like network engineering. The solution that both HP and IBM propose is the creation of package systems that include operations tools tuned not only to the configuration but also to the application/vertical.
Under the covers there aren’t big differences in the technology between the two IT rivals; it’s largely a matter of positioning. IBM is stressing a data-center-in-a-box, something purpose-configured in many cases for key applications. HP is stressing a private-cloud-in-a-box. However, a single data center running virtualization is by anyone’s definition a small private cloud, and a PureFlex infrastructure or PureApplication vertical platforms are both cloud-ready as well. Thus, we’re seeing vendors addressing the operations cost issues in a cloud or no-cloud environment, equally. Which is as it should be; the best strategy for any business is the one that provides the lowest overall cost.
This will probably impact the cloud space, however. First, any standardization in configuration at the data center level makes hybridizing potentially easier, at least for public cloud services that use the same or highly compatible facilities. Second, the real benefit of the cloud has always been pooled operations efficiencies, and this will tend to reduce those benefits by lowering the internal IT cost of the cloud alternatives. And despite the similarity under the covers, I think it’s likely that HP plans to use the cloud more proactively as a differentiator than IBM.
HP’s cloud positioning may be reflected in another HP announcement yesterday, a follow-on to its cloud-talk. HP said it was introducing a series of SOA-based “templates” that would define the operations (deployment and lifecycle) for applications in target vertical markets. If you’ve followed my cloud musings, you know that I’ve always believed that SOA clouds based on PaaS services were the most logical kind to deploy because they could make the distinction between public and private clouds largely one of simple hosting. Further, any applications actually written for the cloud and designed to open new business benefits would likely be componentized, orchestrated, and thus SOA-based. I’m a little surprised at IBM not nailing this particular point down in its own marketing of PureSystems. IBM has slipped a bit in our surveys in strategic credibility, and perhaps they’re not done slipping yet, which could give HP an opportunity. Or Cisco.
HP may see that same dynamic emerging with Cisco, and that could be why one of their first SOA-empowered vertical templates is for the network service provider. Here, HP has gone as far as any player in the market at defining a SOA-cloud architecture as the basis for future services. Since the cloud (public provider infrastructure or enterprise private clouds) has been the primary place were Cisco’s UCS has been successful it makes sense for HP to try to push as many Cisco buttons as it can, and the service layer is a place where Cisco is potentially vulnerable.