Our annual cloud computing survey this year confirmed a trend we noticed last year: adoption of infrastructure-as-a-service is slowing. While we've never thought that cloud computing would zoom to ubiquity in a few short years, we're surprised that IaaS adoption zoomed to about a third of IT shops and then all but stopped. And for the first time this year, expected use of IaaS is actually lower than in previous years.
As we've talked with IT planners, a familiar pattern has formed: IaaS appeals to those with green field opportunities, particularly startups and IT shops that view today's "enterprise" IT offerings as either too restrictive because of licensing terms or inadequate for their purposes. For example, the business models of Web powerhouses such as Google, Facebook, Amazon, and Yahoo work only if they can get computing in its cheapest and rawest form and mold it to do what they need.
Even something as basic as the long-held server vendor practice of including additional sheet metal (and expansion capabilities) as they offer additional CPU and memory capacity is antithetical to the needs of the likes of Google and Facebook. First thing they'll do is rip the sheet metal off, and once they deploy a server, they'll never change or upgrade the physical hardware. They'll use it and replace it. So when Amazon comes along selling the rawest computing commodities, those offerings immediately appeal to the would-be Googles and Facebooks, or to more entrenched companies whose needs aren't being met by the prevailing on-premises models.
[ For more on HP's plans for the cloud, see HP's Cloud Strategy Takes AIM At Amazon. ]
Most of the 33% of respondents to our cloud survey who report using IaaS (up from 31% in 2011) fit those profiles, we think. Of the remaining two thirds, only 27% see no need for IaaS, so that leaves another 40% of IT shops who are trying to figure things out and who generally aren't finding what they need.
Whether or not you buy this version of what's happening in the IaaS market, there's clearly a need for a cloud "for the rest of us." What the rest of us want is a cloud that's not quite such a do-it-yourself experience. Lots of IT managers look at their environments, which support from hundreds to tens of thousands of custom-built (or at least highly customized) applications, and they'd like an easier way to modernize them or a cheaper long-term strategy for maintaining them. They also put a premium on security, performance, reliability, and compliance. None of those things except performance is to be found right now in the offerings from Amazon, Rackspace, and their competitors.
Within that context, you can begin to understand the moves Dell announced earlier this month and Hewlett-Packard announced today. Dell's moves in particular are all about courting IT shops that need help moving their many legacy applications to something more modern.
Over the past couple of weeks, Dell has announced the acquisition of three companies: Clerity Solutions, Make Technologies, and Wyse Technology. Clerity's mission is straightforward: to help manage the migration of mainframe 3270 applications onto something more modern. And if you're going to get rid of those vaunted 3270 terminals and emulators, you'd better have something like a Wyse system to replace it with. In the age of mobile apps and iPads, moving from a 3270 to a Wyse terminal might not seem like much of a step forward, but it could release companies from expensive, outdated legacy systems.
Wyse offers security and control, and for many applications it makes an excellent presentation platform. When companies are ready for that next step, Dell will have the expertise and technology that Make provides to further modernize their applications. And once Dell's Perot systems integrators put the pieces all together, of course they'll want to run the new apps on shiny new Dell servers and storage systems. It's a powerful story and probably well worth the better part of a billion dollars that Dell spent to pull it together.
You'll hear talk about how Dell had to buy Wyse to compete with HP in the market for zero OS systems and secure terminals. That's a load of crap. What Dell is doing is building a portfolio of products that let its customers modernize their application portfolios. Wyse will bring new customers to Dell, and Dell will bring new customers to Wyse--many with needs in these areas.
For its part, HP is gradually releasing details on the worst-kept secret in cloud computing. For more than a year now, HP has been building out data centers to offer cloud services that initially compete with those from Amazon Web Services but will eventually go well beyond--essentially creating a cloud for the rest of us.
HP today introduced a set of IaaS products based on the OpenStack technology. Services will include compute and block storage, as well as a content delivery network offered under a deal with Akamai.
The fact that HP is basing its products on OpenStack is important. Until Amazon's recent announcement of cooperation with Eucalyptus, it was hard to imagine how a customer might operate a hybrid cloud. It seemed that Amazon was pushing an all-public cloud strategy. HP, by embracing OpenStack from the outset, is encouraging a hybrid arrangement from the start--one in which IT shops can run compatible OpenStack systems in their own data centers, making it easier to move applications between their own systems and HP's.
HP doesn't explicitly say that it's gunning for Amazon, but that's what it's doing initially. In an email, HP outlined some sample prices for its compute offering: "Just as an example, pricing for HP Cloud Compute will be based on the instances (XS to XXL) used on an hourly basis ($/hr). For extra-small standard instance type (1 GB RAM; 1 virtual core; 30 GB local disk), customers will pay $0.04 per hour. For double extra-large standard instance type (32 GB RAM; 8 virtual cores; 960 GB local disk), customers will pay $1.28 per hour." That's clearly Amazon-like pricing.
HP plans to add services around databases, identity management, software test and development, and security, as well as higher-level functions based on its Vertica and Autonomy software. These are the kinds of services that will make cloud fence sitters take notice, though HP is vague on exactly how or when those services will come to market.
The devil most certainly will be in the details. Large, well-resourced, and well-run IT organizations--particularly those that view IT as a strategic advantage--will continue to have a hard time seeing the benefits of buying an external service. Those companies will use the cloud tactically. For the rest of us, the right set of cloud services can lead to better-maintained custom apps, but pricing, risk management, and ultimately return on investment will continue to give IT planners pause. While there's a lot to hate about the existing on-premises application development and deployment model, there's 30 years of expertise behind it, and that makes it predictable (even when it's predictably bad).
What IT planners must realize is that vendors see the cloud as a source of new and increased revenue. IDC and Microsoft have predicted that cloud technology will lead to as many as 14 million new jobs, so IT pros who are viewing the cloud strictly as a cost saver are likely be disappointed, even if those job numbers are overstated (vendors have dreams too). What Dell and HP want to do is offer a path to speed up deployments, offering better universal accessibility (inside and outside the corporate walls), performance, security, and availability. And they expect to offer all that for more than what you're now paying for deployments that lack some or all of those things.
By offering systems that appeal to "the rest of us," Dell and HP are on to an important improvement over the current public IaaS cloud model. InformationWeek's survey found that even among cloud proponents, a healthy majority are assigning less than 25% of their total workloads to the cloud. There's a lot of room for adoption.
The pay-as-you go nature of the cloud makes ROI calculation seem easy. It’s not. Also in the new, all-digital Cloud Calculations InformationWeek supplement: Why infrastructure-as-a-service is a bad deal. (Free registration required.)