Big Dogs Weigh in on Cloud: IBM and HP Show the Goods

CIO — Serendipitously, both IBM and HP held events this week to describe their cloud computing initiatives. Their presentations offered insight into what they're doing and provide some food for thought to IT organizations assessing what cloud computing means to their future—as well as some information that might give pause as well.

IBM announced a number of separate things as well as doing an actual demo of cloud capability. First, it's created a Cloud Computing division that reports directly to IBM's head—equivalent to the software, sevices, and hardware divisions. Second, it announced a raft of cloud computing offerings, including services relating to cloud strategy, transitioning current data centers to cloud-enabled data centers, and IBM facilities to enable testing of cloud solutions. Third, it announced a capability to allow IT organizations to use external clouds to migrate workloads from internal cloud data centers to external clouds. The demo showed how an application with multiple systems running software could have some of those systems live migrated to an external cloud. This capability is done through Tivoli, which manages all the systems. Juniper participated in the event, with Juniper networking underlying the application migration with MPLS-based connectivity providing high-bandwidth communication between the internal data center and the public cloud data center.

HP's event was more focused on illustrating HP's approach to cloud computing with presentations by several different groups. HP locates its cloud computing technology efforts within its Technology Solutions Group rather than breaking it out on its own. However, there is a CTO for cloud computing and each group within TSG has cloud offerings. HP also provides a range of services like helping clients develop cloud strategy, creating agile data centers that enable fast resource provisioning and repurposing, etc., as well as hardware and software designed to assist the operation of agile data centers.

So what does this mean for you, the IT practitioner?

First, the two biggest players in IT believe in cloud computing—and that means they believe that it's important to you. While there was some mention of cloud hype at both events, companies like these don't invest hundreds of millions of dollars in something they see as an IT fad. I've seen how big companies like these treat IT trends they view as fads: they put together a mid-level group, buried deep within a main business structure, headed by somebody without a lot of clout within the organization. That person is chartered with going to second-tier conferences and describing a rag-tag bunch of minor initiatives. What IBM and HP presented doesn't have that feel at all. They're dead serious about cloud computing and view it as the next beachhead of computing, akin to the rise of the PC or the arrival of the Internet. If it's important to them, it will be important to you.

Two, for you, in your everyday job, business-as-usual is going to look a lot different. If IBM is setting up a senior organization to push cloud computing, you're going to have to have a cloud strategy. Pooh-poohing it when IBM's CEO is talking it up with your CEO isn't a good idea.

Three, expect a real challenge in figuring out your cloud strategy. In talking with several of the attendees to the events, they mentioned that they got a sense the companies are struggling to present a coherent approach to cloud computing -- it was a "work in progress." My sense is that this is because end user interest in cloud computing is unprecedented in how quickly it's building—one presenter from HP noted that 18 months ago he had to use the term "utility computing" because no customer had heard of cloud computing. I've never seen such a quick growth of end user interest in a new technology offering and it seems even the biggest providers are scrambling to respond. And hey, if they're having trouble in figuring it all out, so will you.

On the other hand, there were some things about the two presentations that might raise concerns.

The demo that IBM showed, while quite impressive, was really constrained in its architecture. It was all open source running on open source Xen boxes—not a typical application architecture. While the controlling management structure (Tivoli) can undoubtedly be extended to interface with VMware and Hyper-V, it seems to me that adding proprietary software into the mix is going to complicate this scenario. I asked about the issues of software licensing and was told that those kinds of issues are already present with in-house virtualization, so this scenario isn't really any different. I never underestimate the ability of software companies to complicate their customers' lives, so I am not convinced this won't be an issue. Furthermore, some people assert that cloud computing requires new application architectures posing issues of enterprise architecture, employee skills, etc., which were elided in this demo. For what it's worth, my opinion is that cloud computing enables system architectures that are quite difficult to implement within the constraints of typical data centers and system architectures, but does not impose a certain style of system architecture; existing system architectures can be successfully implemented in cloud environments.

A bit more concerning is the unmistakable feel of traditional IT about these initiatives. One blog posting I read characterized this as an "enterprisey-feel cloud." The demo—again, quite impressive—requires Tivoli at both ends of the migration; in other words, the external cloud provider needs to have Tivoli in place as well as the originating internal cloud. That means that the choice of external cloud providers is among those that are part of the IBM ecosystem—this was described as being able to use resources that were reserved at an external cloud, which sounds to me like an outsourced arrangement requiring previous contract negotiation, etc., etc., certainly not truly agile, but rather an extension of traditional IT infrastructure arrangements. As another example, I brought up the issue of "skinny straw" connectivity—that is, data center Internet connectivity that may make it hard to handle workload migrations to external clouds; the response was that most data centers have fiber connectivity in place to provide sufficient bandwidth. To be fair, IBM and HP customers number the largest IT users in the world, but I'm not sure that most of the potential cloud users in the world have the resources assumed by these solutions. In any case, I'm not sure that any IT organization is willing to move to cloud computing if it involves being limited to one supplier's cloud resources.

Both IBM and HP proposed moving to the cloud via a more nimble data center, using management tools, governance services, and provisioning software to create an on-demand IT organization incorporating the cloud. I interpreted this as requiring additional investment in order to begin leveraging cloud computing. The whole reason for moving to the cloud is to cut costs (I know, cloud computing also provides the ability to implement whole new classes of systems based on easy scalability, etc., etc.—but the primary motivation is to save money). I'm not sure in this economic environment that a message of spend money now to potentially save money later is that attractive. If I were a CIO—and especially one being chivvied into cloud computing by a CEO hot to trot based on an article in The Economist—I'm not sure I wouldn't say "Let's ringfence investment in the current data center and put some new systems up on Amazon EC2 and see how it goes. We can always come back later to the existing stuff." For sure I'd want to see some proof points before upending my current infrastructure and processes. In my opinion, products and services that address the concerns I've been discussing in my series on "The Case Against Cloud Computing" and offer a prototype capability would be more eagerly embraced.

Overall, it's impressive to see how big vendors are embracing cloud computing and reflects, I think, the ardent interest in the topic on the part of end users. I continue to poll everyone I talk to about cloud computing as to why end users are so interested in the topic; the range of responses are interesting but seem rooted in a dissatisfaction with the current modes and costs of enterprise IT. IT is getting more and more complex, seems to get more expensive all the time despite Moore's Law. and seems unable to keep up with the dramatically changing global economic landscape. Something needs to change, and cloud computing seems like the candidate.

Bernard Golden is CEO of consulting firm HyperStratus, which specializes in virtualization, cloud computing and related issues. He is also the author of "Virtualization for Dummies," the best-selling book on virtualization to date.