Bumpy Ride Ahead for Storage in 2008
For those who don't read Dutch, here is my next column for Storage Magazine in the Netherlands.
Storage 2008: Get Ready for a Bumpy Ride
By Jon William Toigo
In my conversations over the past six months with many senior managers and IT professionals in companies large and small on both sides of the Atlantic, I am hearing a consistent message. 2008 will be driven by what can be called the C-4 objectives: Cost-Containment, Continuity, Compliance and Carbon Footprint Reduction.
A blind man with a cane can see that the economy is getting tight, that folks want to protect the investments they have made in IT, that the regulatory climate is getting a bit more strict, and that – with or without eco-consciousness – electrical power is getting more expensive and more difficult to source in many locations. The C-4 objectives will translate into Front Office policy this year and everyone on the business side of the house, from CEOs to Directors, will be looking at the Back Office, where IT lives, for strategies that will operationalize policy. Looking busy won’t cut it: IT will need to demonstrate its business value in tangible ways.
From where I’m sitting, it all comes down to better management – of infrastructure and of data itself. Storage, especially, will come into the crosshairs because of its increasing claim upon CAPEX and OPEX budgets and because it is mostly un-managed or, in the best of cases, poorly managed. A client recently called to ask me whether he should buy a new array or hire another storage administrator. The answer I felt compelled to provide reflected the current storage conundrum in a nutshell, “If you buy another array, you will probably need to hire another admin to manage it for you.”
Our failure as consumers to demand a consistent and cooperative management scheme from vendors has resulted in (1) extremely poor capacity allocation efficiency in our storage infrastructure (below 10 percent of optimum in many cases) and (2) a requirement to add more capacity on an all-too-frequent basis. As the abysmal implementation record of the Storage Management Initiative-Specification (SMI-S) from the SNIA demonstrates, the industry isn’t going to gift us with a unified management scheme.
SMI might be good for what it portends to offer, but too few vendors are implementing the necessary hooks on their gear that enable it to work. They say it’s because no one is demanding SMI or is willing to pay for it. Closer to the truth: vendors don’t want unified management that makes it a simple matter to unplug vendor A’s gear and replace it with vendor B’s kit.
Unified management, at the end of the day, portends the commoditization of storage: all vendors are selling the same box of Seagate hard disks. In the absence of effective storage management, vendors can sell us proprietary wares that lock in consumers and lock out competitors – and by extension, make heterogeneous storage infrastructure that much more unmanageable.
In 2008, storage will also surpass servers in many shops as the biggest power gluttons in the data center. Our failure to manage (read “classify”) to any significant measure of granularity means that new regulatory requirements that require that certain data be stored for a longer period of time are sure to drive capacity requirements skyward. We can’t readily identify which data is subject to which regulations or directives, so we are storing it all – forever. This, in turn, impacts our response to data protection requirements (both for security and privacy and for continuity and disaster recovery): we back up everything or encrypt everything because we simply don’t know what data is associated with business processes governed by new regs and directives and which data isn’t. Truth be told, we are struggling even to find the data associated with a specific application once it flows into our pools of disk drives.
The go forward strategies for 2008 will vary from company to company, but at root, two things are needed. First, every company needs to choose a storage management “solution” (typically a storage resource management software package) and standardize on it. Then tell hardware vendors that you refuse to buy their stuff unless it can be managed using the package you have selected. This will begin to normalize your infrastructure from a management perspective.
Secondly, we need to start a data management initiative. This will require interviews and investigations with data stakeholders to map stored data back to the business processes that create and use it. This is the essential first step whether you are trying to contain storage cost, identify data assets you need to back up and protect, comply with FID and other regulations, or green your IT operations.
Welcome to 2008. It is going to be a bumpy ride.
Had an interesting discussion today, by the way, with one-time Veritas, one-time SNIA boss, Roger Reich. He left Symantec (and most things storage) last Spring, in part to retry his luck on climbing a Himilayan peak (not sure if he made the summit or not). He has now gotten together with a few of his old crew in the SMI development group at SNIA and launched a new company called Olocity, which aims at bringing about the original vision of SMI (as opposed to the SNIA's perversion). He will be offering consumers and vendors assistance on embedding real SMI providers or taking advantage of them where they are available on gear using some open source technology that he purchased out of pocket (StorageIM). Visit Olocity.com for more insight into his startup.
He is very interested in finding a group of consumer companies that want to build a managed infrastructure using SMI. Interested folks should contact Roger at Olocity, or you can leave word here.