By Mark Peters
There is a natural tendency to look back at what changed during the past year and then make predictions for the forthcoming year. The storage business is no exception. This can be disconcerting for storage professionals who are struggling just to meet their daily obligations and keep systems running at decent service levels-without having the time to get on the leading edge of storage management and technology.
First, let’s take a look at what hasn’t changed over recent years. If you’re worried about not “keeping up” in the short term, it might cheer you up to know that a longer-term view of storage shows that many of the basics have not changed as much as you might think.
What has not changed is more crucial than what has changed. Why? The quick answer is that, for decades and until recently, we’ve lived in a somewhat predictable state. Storage needs have grown rapidly, and the resulting gigabytes, then terabytes, and now petabytes need to be managed, but the storage industry has been very skilful at navigating around the resulting challenges.
Primary data lives on disks. That hasn’t changed, nor has the underlying physics of disk, despite recent advancements. Back then, contention was an issue, and areal densities (capacity) growing far more rapidly than the rpm rate (performance) could easily lead to challenges managing the retrieval of ever-growing volumes of data. We also knew that disks have an uncanny tendency to break. The storage industry has been creative in developing a slew of Band-Aids to mask and manage these issues over the years (i.e., dual pathing, caching, wide striping, and RAID technologies).
What else hasn’t changed? For one, backups have always taken too long. The “Band-Aids” in this arena have included incremental backups, backup to disk, and data de-duplication, which are all approaches to dealing with performance and reliability issues. And then there’s tape-the last bastion for long-term and off-site retention due to its low cost and portability. What hasn’t changed here is that reports of the death of tape continue to be premature; tape still has its place and continues to improve with automation, compression, and virtual tape libraries (VTLs), alongside massive capacity improvements.
And the most obvious unchanging element of all? Storage capacity requirements continue to rise. Today’s demands around unstructured data, rich media, compliance, backup-to-disk, and archive are only accelerating the storage capacity trend. This means that configuring and managing storage keep getting harder, so vendors have found that providing ease of use and management Band-Aids can provide competitive advantages.
It’s reasonable to wonder why the Band-Aid approach continues. It is due to natural human and business inertia (nobody wants to “start from scratch”) and is also because it has worked. Massive surgery can be avoided because simple Band-Aids will do the trick. But can Band-Aids continue to work as the nature of IT and its storage needs change?
There’s one other phenomenon that hasn’t changed over the decades: Every few years, there are new technologies or approaches that go beyond mere evolutionary Band-Aids, although falling short of wholesale storage surgery. To stretch the analogy a little, let’s call them plastic surgeries.
Some have been revealed as more “marchitecture” than miracle (think ILM-a worthy principle, but ultimately no panacea). Some have come and gone and are now re-appearing (solid-state disks and virtualization have been around for years). Some drive a new facet into the evaluation of technology (the “green” focus is driving innovation and a renewed focus on TCO). Finally, there are the few, truly unique approaches that are making a difference today (good examples are thin provisioning and MAID, which overturn established storage “rules”).
These technologies and trends demonstrate that the industry today remains focused on both the more- efficient use of storage and the use of more-standardized components.
New challenges, new thinking
What has always remained the same is the nature of storage itself and storage management. This storage status quo has existed because the vast majority of work performed in data centers has been transactional. Yes, the task gets bigger and more complicated, but it remains under the control of the organization itself. And it is change in this dynamic that is going to drive the need for a long-term change in storage.
As we move to a non-transactional world (certainly in terms of capacity), a different approach will become necessary. Mix in the impact of Web 2.0 on storage, and the need for a different approach is greatly amplified. As data types and demands finally change, storage managers will need a different way of thinking, rather than just using Band-Aids and plastic surgery. It’s not yet clear what the “next big thing” in storage is going to be, but it is clear that it is going to be both information-centric and highly dynamic.
The storage pyramid
The “storage pyramid” has also escaped change-for the most part. There are high-speed, low-capacity, high-cost options at the apex of the triangle, moving gradually down through a scale of technologies with increasing capacity and decreasing performance and cost. It always was, and still is, mostly about money. Trade-offs like performance and capacity wouldn’t be needed if everything were available instantly for next to no cost. For storage, the last few decades have brought improvements to the pyramid by adding more “steps” that “smooth” the sides and make the technology jumps less abrupt. But it’s still a hierarchy that requires considerable management and relies upon IT professionals to have a clear idea of, and control over, what they need to store. The new world of expanding non-transactional data changes everything.
In a Web 2.0 world, the economics of storage have to change, and, even more importantly, the ability for storage to scale dynamically and automatically in any direction (performance, capacity, throughput, protection, etc.) becomes crucial. The storage pyramid, controlled and implemented in data centers, needs to be replaced by a malleable, amorphous, massive storage “blob,” controlled by and reacting to the varying demands of users. This will break the Band-Aid and plastic surgery mold and require revolutionary changes.
Mark Peters is an analyst with the Enterprise Strategy Group (www.enterprisestrategygroup.com).
White Papers ESG Lab Review Series
ESG Lab Review Series