Declaration of Independence, Storage Style
Given the nature of technology and of business economics, are we ready for a storage revolution?
The July 4 holiday may have passed, but the timeless themes which we revisit during the celebrations at my house by re-reading the Declaration of Independence aloud continue to ring in my ears. If you haven’t read the document recently, you might think of doing so. It establishes, metaphorically at least, the grounds for changing your relationship with your storage vendor.
The truth is that, for reasons of convenience or perhaps even business savvy, many firms have joined their infrastructure at the hip to a specific vendor in recent years. At some point, it may have seemed appropriate to establish a relationship with XYZ vendor for both technical and procedural reasons. Perhaps the vendor was offering best-of-breed storage systems that met the needs of an application -- or the preponderance of applications -- that we fielded in our shops. From a planning and purchasing perspective, it may have seemed easier to go for the one-stop-shop/one-throat-to-choke model by sourcing everything from a single vendor.
However, in the course of human events, as the Founders might say, sometimes you need to reconsider your options. Given the nature of technology and of business economics, now may well be that time.
On an increasing number of sites that I visit, and in an increasing number of conversations with many business and technology professionals, disgruntlement with the status quo is surfacing. One IT manager asked recently why he needed to pay for 100 percent of the price tag for “value-add software” delivered with his storage system when he used less than 10 percent of the functionality offered. Why indeed?
Another IT maven in a different company, noting that disk drives this year are half as expensive on a per-GB basis as the drives offered last year, wonders why the prices of new arrays (collections of disks) double what they were last year? A good question.
Still another complained that he would love to buy the right gear to support his apps, but his options have been predefined for him. He can only source from a list of approved vendors on a list prepared in the front office -- without giving him a vote.
Add to this list of issues some older ones that have begun to resurface of late. Network Appliance, now NetApp, is pursuing gray-market vendors and used-equipment resellers of its older gear with a vengeance. The company insists that licenses for software that gives their Filers their smarts are not transferable with the tin cabinets and spinning rust, so when you take a Filer off the line, it becomes a boat anchor. Despite the fact that leasing companies get their gear (out of necessity) with transferable licenses, then recoup their investment by reselling used gear to the equipment brokers, NetApp’s CEO recently stated that no transfer of licenses has ever been sanctioned by NetApp. He has publicly instructed his corporate attorney to go after anyone who tries to resell a used box -- even those that had transferable licenses to begin with.
We have also seen many smart ideas from smaller players disappear altogether or be placed into proprietary platforms by three-letter acronyms that are busily gobbling up the small fry. XIV is an example. IBM bought the company a few months ago (shortly after we wrote very positive things about them here) and is determined to stove-pipe the technology only on a particular array that will only be sold by IBM.
A Coming Rebellion?
The list of issues goes on. Metaphorically, we have most of the same conditions today that the Founders faced in their arguments with England in 1776. In the face of the current economic slowdown, the requirement to cut CAPEX and OPEX requests while delivering improved service levels, and the mounting costs of storage technology itself, I am seeing an increased desire to rebel in many consumer shops.
In my mind, rebellion equals getting back to the practice of purpose-building storage infrastructure. That means building storage to meet the specific needs of applications, buying only what you need, and making the best of what you already own. Here are the ten tenets of a purpose-built strategy for storage independence.
First, let us take as a self-evident truth that one-size-fits-most storage products never fit anyone's needs very well. Small shops may do well with an all-in-one-box configuration, but not enterprises with hundreds of applications. As your company grows, its technology-support requirements become more complex and the data coming out of different apps need different resources and services.
Second, to understand data hosting and service requirements, we need to examine business processes and applications themselves. Such an analysis yields all of the information needed to develop a set of storage requirements to guide purchasing. The way many companies do things today (encouraged by leading storage vendors) -- fitting applications to infrastructure after infrastructure is built -- almost never provides a good fit of technology to data or to the business.
Third, the core elements of storage are the same: spinning rust, spinning plastic, or streaming mylar. These are properly regarded as commodity components, regardless of what your vendor says. These commodity components are used in everyone’s finished arrays. We need to find ways to leverage the commodity pricing of commodity wares to drive down the cost of finished arrays to become cost-efficient in storage investments.
Fourth, the storage industry isn’t going to help much with respect to number three above. The industry cobbles commodity components into finished products that join proprietary "value-add" software at the hip to proprietary controllers in order to lock in the customer and lock out the competition. (This is an actual quote of a former EMC CEO from a presentation he made in Santa Barbara, California in 2001 before a group of venture capitalists and investment bankers. I’m not making this up.) This creates stovepipes of data that make sensible data management, capacity administration, and cross-platform management very difficult, and drives up storage cost of ownership.
Fifth, purpose-building storage requires the deconstruction of array "systems." Some services provided on array today need to become services deployed in a network so that they can be extended to all storage platforms, not isolated to select storage silos. Xiotech understands this with its ISE platform, which is enabled with features that allow a variety of third-party vendors to add value around the disk subsystem -- so you can buy the features you need and add more as your needs change over time. There is no shortage of network-based service hosting platforms -- from Crossroads Systems’ router gateways to appliances -- where “value-add” functionality can be hosted so that data from numerous applications can be exposed to the services en route to their target storage locations.
Sixth, to make purpose-built storage work, cross-platform management must be a forethought, not an afterthought. In many shops, storage is still managed on a one-off basis, using the device manager that came with the box. Enterprise-level management requires a more holistic, cross-platform approach. I like developments in the realm of Web services that portends connecting applications to network-based storage resources and services. Until a comprehensive Web services model is created, however, there are very good SRM tools available in today's market.
Seventh, management plus purpose-built storage can drive substantial cost out of storage technology, which currently consumes between 30 and 70 cents of every dollar spent on IT hardware and accounts for the lion's share of electricity used to power IT infrastructure.
Eighth, management can and should be used as a selection criterion for storage investments. If the gear being proposed by a vendor cannot be managed using the storage management tools adopted as a corporate standard by your shop, don't buy it. This compels vendors to cooperate and to support a common storage management approach -- something that SNIA's Storage Management Initiative has utterly failed to do.
Ninth, purpose-building storage goes hand in hand with data management and intelligent archiving to drive out cost and improve both allocation and utilization efficiency in storage investments. Intelligent archiving (preferably to tape or optical) is a strategic play that every competent CIO should be building today. Without it, you are growing disk capacity at a budget-breaking rate.
Tenth, purpose-built storage is going to require IT to do some of the integration they currently outsource to vendors. It lacks the appeal of the "one-stop-shop" or "one-throat-to-choke" approach popular in recent years, but it also eliminates vendor lock-in, overpayment for value-add features that you don't use, and a lack of common management that increases OPEX costs for storage.
As with the American Revolution, it took nearly a decade to throw off the old system completely and adopt a new way of doing things. Since then, the American experience has been a work in progress. The same will be true of storage. We can’t rip and replace everything overnight, but must phase in change over time. The point is to declare independence today and to establish a vision to work toward in the years to come. Replacing stovepipe systems with open storage platforms and network-based storage services, implementing Web services-based management, and deploying an intelligent archive are ingredients of the vision and should guide purchases from this point forward.
That’s what I’m telling my clients. Your comments are welcome. firstname.lastname@example.org.