In-Depth

Aligning Storage Costs with Business Priorities

A proposal for a “three dimensional approach”: automating storage provisioning and device management, data movement, and data replication.

• By Jon William Toigo
• 03/04/2004

When mail arrives from either of these guys, I tend to open it immediately. Rob recently sent me an e-mail that struck me as a kind of manifesto for clear-headed storage architecture. I have chosen to reprint much of it here.

Nieboer begins with insights about the kind of storage planning companies require. Citing the failure of FC SANs to deliver on their business value proposition, he says the time has come to stop looking for panaceas and solve problems for ourselves.

“Clearly, there is a need for people to think strategically about storage. I have been urging the creation of an Information Strategist function within IT, and this person may very well also be a Storage Architect. These people ought to be thinking about the information needs of the business for next year, 3 years from now, 5 years from now, 10 years from now. The architect ought to translate these needs into some fundamental architectural building blocks that can form the foundation of a sustainable storage infrastructure that supports business goals and objectives, and which aligns storage costs with business priorities and value.”

Based on years of observation and work with client companies, he concludes, “Today, I believe many (most) IT shops have a broken, unsustainable storage model. Information is growing faster than disk cost is declining, and faster than our ability to manage storage is improving. All this is happening when IT budgets are flat to declining, and while information has never been more critical to the survival of the business.”

Tactical storage hardware purchases are a major component of the storage cost explosion today, says Nieboer. “The tactical behaviors of storage managers are bringing storage in these shops to the brink of breaking. They run out of storage, they buy some more of what they have, or what's cheapest. Yet, every new storage decision simply adds to the problems they already have, rather than begin[ning] to set the foundation for a sustainable infrastructure.”

He calls for a new alignment of “storage costs with business priorities” and suggests that a “three dimensional approach” is required: we must automate storage provisioning and device management, data movement, and data replication.

First, organizations need to attack the labor costs associated with storage provisioning, he says, who clicks off the key questions to be answered almost by rote: “Where do I store the first copy of the data? What performance characteristics does this data need? How long should it stay here before I move it somewhere else? How and with what human resources and with what software do I manage it?”

Nieboer offers a straightforward answer: “Find every avenue you can to automate storage management! Act opportunistically and tactically and seize upon tools and technologies that automate various management functions.”

He cautions against a silver bullet mentality, especially as it pertains to the multi-tiered architectures being advanced by most storage vendors today. “Clearly, there are some compelling reasons why we should exploit multiple tiers of storage. The cost benefits are clearly there. The only caveat is that you had better find an automated way of managing the movement of data across a storage hierarchy, because we don't have enough people to do this.”

Second, Nieboer urges the automation of data movements wherever possible, understanding that tactical selection and deployment of automation technology may lead to some dead ends and the requirement “to abandon investments a year or two down the track.”

Nieboer’s third target is data replication, which he says “is becoming the real craft behind the broader backup/recovery and business continuation problems.” Technologists have been largely remiss in analyzing how many data copies are needed, whether copies need to be hosted on the same type of platform as originals or can be adequately handled by cheaper Serial ATA platforms, and how much data needs to be replicated synchronously versus asynchronously. He says there is little attention being paid to addressing the nagging issue of how best to leverage “the absolute low cost of tape for both long-term retention of backups and for the archiving of data which is under some kind of regulatory oversight, while eliminating the impact on applications and application servers of potential mechanical failures.”

Adding intelligent self-management features and virtualization capabilities to tape platforms is inevitable, in Nieboer’s view, to eliminate the direct connection of applications to tape technology, to address the data restore requirements of contemporary business continuity requirements, and to compensate for the data management shortcomings of technologies such as FC SANs.

Treading a fine line between his beliefs and his employer’s line, Nieboer summarizes, “[The industry] has created two intersecting worlds. The first is the world created by the application servers and the buffered appliance. In this world, no physical tape exists, backups almost always work, and recoveries are fast and predictable. The second world is the one created by the buffered appliance and real backend tape. In this world, which operates independently of, and asynchronously with, the application servers, data are migrated from the disk buffer to tape under policy, maybe with multiple copies both local and remote, data may be recalled back to the disk buffer, and outboard migration can occur independently of application servers from older tape technologies to newer higher capacity technologies as a background task.”

If companies adopt this unassuming outline of strategic planning goals, they may well begin to bring their storage challenges (and budgets) into alignment with business priorities. Then, “Bob’s your uncle”—as one might say in Sydney.

What do you think? E-mail us at jtoigo@intent.net.