The Role of Storage in IBM’s New PureSystems

Paraphrasing Frank Sinatra’s famous "Love and Marriage" (lyrics by Sammy Cahn and music by Jimmy Van Heusen), we might sing: Systems and storage, systems and storage / Go together like a horse and carriage / This I tell ya, brother, you can't have one without the other. In this context, IBM’s new PureSystems strengthen an already strong bond between the company’s system "horse," which does the CPU pulling, and the storage "carriage," which carries the information payload.

Overview of IBM PureSystems
Our focus here is on how IBM storage works with IBM PureSystems. But first, for those of you who are unfamiliar with IBM PureSystems, a brief introduction is in order. For those of you familiar with the product family, a quick recall should be helpful. IBM recently introduced what it calls a new breed of solutions under a concept that it calls "expert integrated systems." IBM states that such systems combine the flexibility of a general-purpose system, the elasticity of cloud and the simplicity of an appliance. The twin goals are to fundamentally change the experience and economics of IT. By experience, IBM really means driving out the complexity associated with managing an information infrastructure, and by economics, the company is really talking about squeezing costs out without compromising service-related metrics.

Those goals are nothing new for the IT industry, and a lot of progress has been made over the years, but with all the changes in the IT world, including server virtualization and the cloud, the battle to eliminate complexity and reduce costs seems to be never-ending. IBM backed up its PureSystems talk with an investment of $2 billion in R&D and acquisitions over four years. The company is also leveraging a broad ecosystem with applications from about 125 ISVs that are certified as "PureSystem Ready."

IBM defines three pillars for expert integrated systems: built-in expertise, integration by design and simplified experience. Built-in expertise is about capturing and automating what experts do, from the infrastructure to the application. Integration by design is more deeply integrating and tuning hardware and software. Simplified experience yields are integrated management of the entire system.

The first two PureSystems product fruits are the IBM PureFlex System, which focuses on integrating and optimizing system infrastructures, and the IBM PureApplication System, which takes an application-aware approach to the system for what IBM expects to deliver rapid time-to-value.

And now on to the storage carriage.

How Storage Works with IBM PureSystems
Servers today are almost always discussed in the context of virtualization, as that is the way IT infrastructures are headed. Storage is also headed in this direction; in IBM’s case, it uses its storage hypervisor to virtualize storage. Now, if you have the opportunity, please see my earlier piece on IBM Pulse 2012: A New Storage Hypervisor. That piece describes how the IBM storage hypervisor virtualizes storage.

Two of the key benefits of using a storage hypervisor to virtualize storage are:

--Improve storage efficiency: Improving the storage provisioning process by not over-allocating storage up-front leads to higher (and therefore better) percentages of storage utilization; enabling different service levels for different data types (say, email and OLTP applications) improves the use of I/O resources (letting them do more before upgrading is needed) and also avoids making more copies than necessary for data protection purposes. All in all, managing physical storage resources better yields an economic cost benefit in better (dare we say optimal) use of storage assets; that ties in nicely to the change of IT economics goal of IBM PureSystems.

--Improve the manageability of storage: This occurs both through an inner and an outer focus. The former makes storage provisioning faster, easier and more flexible. For example, long-range planning for application storage requirements is unlikely to be accurate, but traditional provisioning to either reduce or expand the allocated space for an application is difficult and most likely results in painful application downtime; in the brave new world, logical allocations can be easily set and changed without real physical world negative impacts. The latter (outer) focus is the ability to manage heterogeneous arrays (from non-IBM manufacturers), if desired, as one grand managed pool of storage; that extends both the efficiency and manageability to much more than just a single array.

Now, IBM’s storage hypervisor has a heart of IBM Storage Volume Controller (SVC) and IBM Tivoli Storage Productivity Center (TPC). The Storwize V7000 runs the same software at its core. Note that TPC is part of IBM System Director, which is part of the IBM PureSystems management environment. Thus, integrating IBM PureSystems and the IBM Storwize V7000 essentially means that you have the storage hypervisor for use in an IBM PureSystems environment.

However, these benefits are achieved only when running the storage hypervisor in conjunction with a Storwize V7000 storage system. Of course, all the rest of IBM’s storage portfolio, including enterprise systems storage (the DS8000 family and the XIV family) and file systems storage (including Scale-out NAS and the N series), are compatible with IBM PureSystems (which means that you can run them with storage business as usual), but they are not optimized.

But, wait! There is a way out of this dilemma. Simply put, place a Storwize V7000 system as the front-end focal point within a set of storage arrays and let it provide the overall management framework for turning IBM’s other storage systems and any of the 100-plus non-IBM arrays qualified for IBM SVC into virtualized storage, while at the same time taking advantage of the performance, availability and scalability capabilities of these other arrays.

Mesabi Musings
In the PureSystems launch, we saw the opening salvo in IBM’s use of expert integrated systems to reduce IT complexity and reduce costs. Whether we call it a bold move toward transforming the data center, creating the infrastructure for a true cloud (one that is truly elastic) or IT-as-a-service, it will require a lot more analysis and attention as time goes on.

Note that IBM’s competitors are not standing idly by but are also promoting their own initiatives, and probably preparing their big guns for a salvo. Now, whether all this is truly revolutionary or simply evolutionary (but with positive benefits) will have to be determined after the fact. Stay tuned--however it turns out, change continues to be in the air.

IBM is a current client of David Hill and the Mesabi Group.