Intelligent Enterprise

Better Insight for Business Decisions

RSS
Webcasts
Whitepapers
Subscribe
Home

August 12, 2002

The Next IT Imperative

As storage investments grow, sound information infrastructure will rely heavily on a solid storage strategy

by Richard Scannell

Continued from Page 1

For example, with that approach in mind, what backup model would be appropriate for retained data vs. affecting data? For retained data, perhaps tape backup with duplicated tapes to an off-site location. For service-affecting data, perhaps a local disk copy with a remote disk copy and a tape copy of the remote disk, with duplicates to an off-site location.

It's difficult to complete the exercise without going to your vendors for values to plug-in to your model, but by understanding the required attributes at each data classification you can clearly articulate the cost model and associated benefits at each level. In organizations that charge-back to customers this data is invaluable, particularly as the price of raw disk drops lower and lower. The ability to articulate the cost model for a given class of storage quickly disarms the argument of why IT charges so much for storage vs. what it costs on the street. Some attributes to consider include availability, backup, management, and connectivity models.

Developing Organizational Confidence

Now that you're freshly armed with a new sense of what's needed and how the data should be laid out, you'd think the next logical step would be to start implementing technology. Not so! Most storage spending derives not from capital acquisitions, but rather from the cost of managing these assets. Therefore, it's essential that you ensure the organization responsible for supporting this newly aligned strategic element of IT is appropriately prepared for the tasks at hand.

To support the never-ending growth of data, companies will be forced to transition from a world of distributed storage to one of centrally managed storage. As that transition occurs, many of the processes and policies that once were appropriate and effective will not scale to the task at hand. Take two simple examples: change control and communication. In the world of distributed storage, where an administrator manages a set of devices, each time a change is required (a patch), the only individuals who need to be notified are those associated with that device — the largest risk being that the given device will be out of commission for some period of time, affecting only those users. However, in the new world of centralized storage, the paradigm of individuals managing devices breaks down. Now, think of administrators managing data sets across shared devices. Accordingly, any intrusive change must be communicated to everyone associated with the macro system. Because many administrators are probably involved in managing the macro system, the need for documented change control, test plans, and rollback plans are dramatically higher. If the macro system goes down, all the users are affected.

One of the common fallacies about moving to centralized storage is that it's a technology problem. While not wishing to understate the technology issues, successful implementation of the correct technology is only one of a series of critical success factors in centralizing storage. Taking into account the level of training and competency development that occurred around CRM and ERP initiatives, you must reconsider what will be necessary for the organization to support a centralized storage model. Once again, it's not about the vendor products — it's about policy development, process definition, metrics reporting, communication channels, and user management.

Implementation

At last, we've arrived at the point of technology implementation. It's no accident that selecting and implementing technologies is one of the later steps in this process. Beware of vendors that will conduct a "free" assessment right out of the gate: When was the last time a vendor did a free assessment and recommended another vendor's product? When considering replacing a car vs. fixing one, you typically approach a mechanic who doesn't sell cars in order to gain an independent, nonbiased opinion.

A starting premise for any storage strategy should be a stated goal for reusing existing assets. By adopting a data classification and attribute-based architecture model, reusing existing assets is possible. An asset can be reused based on two factors: the ability to meet a given attribute-based infrastructure standard and positive impact on the cost model. New technology can be selected around the strict guidelines of the attribute-based architecture for each given class of data. Consequently, you can develop far more strict and intelligent request for proposals, in direct contrast to traditional approaches that make storage a footnote to a project.

Measuring and Reporting

Finally, the issue of measuring and reporting must be tackled. The obligatory charts documenting availability, throughput, and backup windows will be needed. But what about the measurements that indicate that costs and quality are kept in line with the plan? For example: How much does it cost to back up a terabyte? How does that differ from service-affecting data to retained data? And how does that change over time as tape utilization, backup leveling, and other qualitative factors begin to decline in efficiency?

Rate This Article

Comments:

Optional e-mail address:

If the only measurements you're taking are Boolean — backups are being completed or they're not — then the best you can hope for is inefficient but functional operations. Far more interesting is this question: At what cost do backups run, and how can that cost be rationalized without sacrificing quality of operation? By considering the next level of metrics around storage operations, you will not only be clearly ahead of your executive team in thinking but also ahead of an industry that, until recently, primarily focused on increasing the capacity of a disk drive and pushing to the next standard byte increment, from megabyte to gigabyte to terabyte and beyond.

It's well known that there's an incremental cost associated with change that's only eliminated when the change is completed — the so-called jumping the gap problem. The goal must be to get from the start of the change to the end of the change as efficiently as possible. By developing a storage strategy, a plan of attack to jump that gap will be set. Because the growth of installed storage has been so fast and the rate of acceleration so high, organizations must move quickly to prepare themselves to manage the situation or they'll find themselves with vast quantities of storage and out-of-control budgets, where that gap will become wider and wider.

The recent economic shift has motivated CIOs to become even more cost conscious. Are there hidden cost savings lurking just below the surface of a current storage architecture that's inefficient, unwieldy, and in many cases, broken? Does it make sense to develop a strategy around storage, if only to get to the inventory and data classification stages so that at least the magnitude of the problem is understood? What reports are needed to effectively measure the problem, and can IT articulate why the current architecture costs what it does? If the questions are many and the answers are few, consider putting storage near the top of this year's list of "initiatives."

Richard Scannell [rscannell@glasshousetech.com] is cofounder and vice president of corporate development and strategy for GlassHouse Technologies Inc. (www.glasshousetech.com), a vendor-independent storage strategy and services firm.