Big Iron in the Clouds, Revisited
The mainframe already embodies a viable proof-of-concept for cloud computing in the enterprise. Is it a cloud platform par excellence?
There's been considerable debate about when cloud computing will be ready for prime time in the enterprise.
The controversy doesn't so much concern the usefulness or broad applicability of the cloud model – consider the many cloud use cases (from prototyping and proof-of-concepts to quick-and-dirty deployments and development test beds) -- as it does the feasibility of an enterprise-friendly refinement of that model: namely, the concept of the private cloud, or of the cloud deployed inside the enterprise and managed by IT.
When (or if) a private, enterprise-focused cloud vision does begin to take shape, IBM Corp.'s System z mainframes will be well-positioned to reap the benefits. That's due in part to the efforts of IBM itself and the vision and zeal of System z enthusiasts, who -- in several industry forums -- have been talking up enterprise clouds based on System z for more than a year. Some experts argue that the concept of an enterprise-friendly private cloud could begin to gain traction precisely because of the mainframe's suitability -- as a centralized, highly-virtualized, multi-platform cloud host -- and because Big Iron already functions as a kind of enterprise-in-the-clouds proof of concept.
Part of the explanation has to do with Big Iron's resurgence.
"The centralized style of computing that the mainframe pioneered is back in fashion. Cheap hardware, deployed willy-nilly, turned out to have unobvious costs. It tended to have very low utilization, was hard to manage and update, and security was a big problem," observes Gordon Haff, a principal IT advisor with consultancy Illuminata, in a recent report.
"As a result, even though scale-out x86 hardware is still very much in vogue, it's increasingly administered as a centralized resource under the control of IT. Cloud computing in its various forms takes this recentralization to the next level." It's in this respect, Haff continues, that the venerable mainframe has become "an exemplar" of what cloud computing in the enterprise should look like. "The 'volume computing' world now covets traditional mainframe virtues such as dynamic workload management, high levels of reliability and availability, and the ability to handle vast I/O loads. As a result, today's System z mainframe looks not so much like a relic of the past but [as] an exemplar of where much of computing is going," he points out.
System z proponents have been touting Big Iron as a cloud computing exemplar for a while now. Take Jim Michael, a vice-president with the mainframe user group SHARE and an associate director at the California State University in Fresno. Michael has championed a vision in which -- far from being a mere use case -- cloud computing will actually drive mainframe usage or adoption inside of organizations.
In the past, after all, customers always knew that they were dealing with a mainframe -- with the mainframe, in line-of-business speak. The enterprise cloud alters that paradigm because it abstracts Big Iron's hardware and software nuts-and-bolts nature -- i.e., its perceived quirkiness. From an internal customer's perspective, he argued, the enterprise cloud is (or should be) a platform-irrelevant utility, such that line-of-business customers can purchase capacity at fixed price points and for agreed-upon service levels in much the same way they have been licensing mainframe capacity for decades. The difference is that the mainframe-based cloud is a much more customizable -- if not completely user self-serviceable -- proposition.
"You could have this model in which [the user says,] 'This is the OS I need. These are the requirements I have. Make me a server instance,'" explained Michael, in an interview ahead of the March SHARE show. "This is going to let us be much more nimble in meeting the user's requirements because we don't have to go through the pyramid cycle to bring in the server or storage resources they need. We're architecting to bring these out of our aggregated resources."
A Not-So-Radical Proposition
What's so intriguing about the mainframe-based cloud -- and what simultaneously plays to Big Iron's strengths as a platform-par-excellence for cloud computing in the enterprise -- is that the cloud model isn't as radical a paradigm shift as many like to make it.
According to Haff, IT organizations are already heavily invested in existing -- and typically tightly coupled -- "legacy" technologies. For many reasons, these technologies can't be wholly transplanted from the distributed space into the clouds. A viable consideration of the private cloud must accommodate the basic hard realities of computing in the enterprise: i.e., the requirements that legacy applications or technologies need to function normally (including, for example, low-latency connectivity between DASD arrays and the server hosts to which they're tethered.)
"[N]either existing enterprise requirements nor the software created to address them are going away -- nor should they. Therefore, an enterprise cloud computing strategy isn't about wholesale rip and replace but about integrating new applications and operational practices with existing ones and orchestrating them in a coherent way. Integration and orchestration are two areas where System z has a track record," Haff points out.
So does Big Blue, for that matter: "IBM … approaches cloud in an enterprise context as essentially an evolution of enterprise software and processes into the cloud, as opposed to an evolution of Web 2.0 into the enterprise. This view of the cloud is through the top-down lens of an enterprise architect in contrast to the bottom-up, tactical, always-in-beta methodology that's been most associated with the consumer cloud."
There's another wrinkle here, too, according to Haff: Big Blue's System z-centric take on cloud computing isn't a radical departure from (or an imaginative recasting of) its On Demand strategy. Instead, Haff suggests, it's an extension of IBM's efforts in the On Demand space.
"The conceptual similarities [between the two strategies] are there in attributes such as application-level availability and resiliency, policy-based automation, high levels of security, and end-to-end systems management," he observes. "What is new -- or at least taken to the next level -- is an emphasis on a heterogeneous dynamic infrastructure of which System z is just a part, albeit the part in the middle of things."
Big Iron Pipedream?
Skeptics might object that proponents of the Big Iron cloud are not realistic. Few, if any, shops are going to purchase and deploy mainframe hardware (to say nothing of hiring mainframe sysops or programmers) chiefly to pursue cloud computing strategies.
Skeptics might also point to the results of a recent SHARE survey, which painted a dismal -- or decidedly non-cloudy -- picture of mainframe cloud uptake. Just 13 percent of large shops (and 10 percent of small and midsize enterprise customers) are using mainframe-based clouds. On the other hand, SHARE officials posit a correlation of sorts between the use of advanced virtualization technologies (e.g., storage and network virtualization) and cloud adoption.
"It is notable that cloud computing is a logical extension of more advanced internal virtualization efforts. Respondents that are moving to cloud computing already have a greater tendency to have other forms of virtualization in production, the survey reveals," wrote Joe McKendrick, the author of the SHARE survey and an industry veteran who has extensively covered the mainframe and minicomputer markets. For example, McKendrick notes, just over half of cloud adopters "are already engaged in application virtualization -- versus 16 percent of the entire survey group."
SHARE's Michael rejects the idea of a pipedream. Big Iron cloud boosters are all too cognizant of both the economics and the physics of the mainframe marketplace, he counters. They've been operating within these constraints -- first surviving and then thriving -- for almost two decades, after all.
In this respect, Michael positions the Big Iron cloud push as analogous to other more-bang-for-your-mainframe-buck initiatives, such as Linux on System z or Big Blue's more recent special engine offerings (e.g., zSeries Application Assist Processor or zSeries Integrated Information Processor). It isn't so much an issue of generating significant new traction for the mainframe -- although that will likely happen, at least to some extent (in accounts in which mainframe activity has been depressed; in former mainframe accounts; and, to a limited degree, in newer shops); instead, it's an issue of investment protection by the continuous expansion of the mainframe's role in (and strategic importance to) an organization.
For this reason, cloud computing is just one of several related irons in the mainframe fire. At March SHARE, for example, a big theme was "high-density computing" -- a deceptively simple term that broadly describes an enterprise future in which virtualization, cloud computing, and other extremely dense, utility-like services predominate. Because of its history and maturity, the mainframe will have an important role in any such enterprise of the future. What's more, Michael concludes, its value-add as a platform for cloud-based computing seems like a no-brainer, too.
"[The mainframe is] not a panacea, but is it a player? Yes. Is it a big player? Yes. [Today] when you're looking at organizations that are offering cloud computing external to the enterprise, outsourcing to some extent, or rightsourcing some of your workloads into the cloud, there are people using mainframes to offer these services," argued Michael, in an interview prior to the March SHARE conference.
He cites a use-case that his own shop at Cal State Fresno is pursuing. "One way I can deal with capacity requirements is to create centers of capability that function as a cloud to end-user departments that need to take advantage of the compute capacity but don't need to worry about acquiring the hardware or software. That's the most interesting enterprise use case, if you ask me."