The need for a coherent view of the storage infrastructure and advance warning of volumes approaching capacity hasn't disappeared. And point management products that come with specific arrays simply can't do the job of an SRM (storage resource management) system across today's complex enterprise storage systems. Yet, despite new efforts to deliver an open SRM standard workable across vendor products, including arrays, switches, tape libraries and optical jukeboxes, analysts peg SRM penetration at just 5 percent to 20 percent of its target audience.

The explanation may well come down to a combination of industry politics and market forces, as tactical concerns and thin budgets move SRM to the back burner.

"We just don't have the time or resources to focus on deploying an SRM package," says an IT manager at Blue Cross Blue Shield. "Compliance initiatives and putting out daily storage fires keep us all busy."

SRM also may be such a low-level "techie" concern that an effective business value case isn't being made in the front office. Vendors have focused on selling wares to solve the problems of data retention (archiving) for regulatory requirements, rather than tools for disk grooming. The end result: no bucks, no Buck Rogers.

Vendor marketing and industry politics have played a role, too: Everyone from big iron purveyors such as EMC, Hewlett-Packard, Hitachi Data Systems and IBM to middle-market frame makers such as Compellent Technologies, 3Par and Xiotech are centering their sales pitches around on-board SRM software that will make all storage manageable--as long as you stick exclusively to the vendor's gear. Storage array-based SRM is a "value-add," intended to prop up the price for otherwise commodity hardware.

For those consumers with infrastructure from different suppliers, the value of third-party SRM tools--such as CA BrightStor, Sun Microsystems' Business Analytics, Tek-Tools' Storage Profiler and others--is being thrown into question by the SNIA (Storage Networking Industry Association), which is touting its SMI-S (Storage Management Initiative), and by new rival Aperi, a standards-based, open-source framework for storage management, launched last year by IBM and others.

Industry efforts to make standards-based heterogeneous storage management a reality have been tediously slow and have been described by insiders as mired in complexity and bureaucratic infighting. Although individual vendors have made announcements that they're adopting SMI-S, a closer examination reveals that SMI interfaces are written onto platforms only as a response to a customer's request and only when the customer is willing to pay extra.

Truth be told, it's not in the interest of the vendor community to enable cross-platform management--that would mean eliminating differentiators between boxes of commodity disks and making it easier to replace brand X gear with brand Y.

To collect information about array status and learn what little they can about actual disk utilization and array performance, platform-neutral management software vendors must go to each equipment provider in turn and beg for, or buy access to, proprietary APIs, SNMP MIBs or command-line interfaces. The result is uneven management performance across platforms.

"We are a predominantly IBM shop," the Blue Cross Blue Shield manager says. "But even with its tools, we can't collect the data we really need to manage our infrastructure. Tools like Tivoli Storage Manager and Storage Tank components provide a lot of information--piles of information. But none of it is relevant."

A storage manager at a large cell phone company agrees. "We bought EMC in part because ECC, its management tool, was supposed to manage all our gear. It did a fair job with our DMX arrays, but couldn't manage other frames very well, including EMC's own Clarion products, let alone third-party gear," he says. "We brought in AppIQ; same dog, different set of fleas. Veritas, same story. It would be great if we could take the pieces out of each SRM suite that actually work and cobble them together. If you could do that, you might have a real story to tell."

An Open Secret

Industry insiders confirmed these views. "There have been no giant leaps in storage management in the last year," says Anders Lofgren, CA's senior vice president for product management and strategy. "SNIA still has a long way to go to fulfill the promise of SMI, so we're still relying, in part, on proprietary interfaces to collect information about underlying infrastructure. We don't foresee that changing in the near- or midterm."

Lofgren adds that going with homogeneous infrastructure is not the SRM alternative that hardware manufactures make it out to be. From a business perspective, he says, it just isn't realistic to believe that any vendor has all the products a company needs. SRM, he concludes, is the purview of the hardware-agnostic software maker.

A somewhat contrarian position is offered by Mike Riley, a solutions architect at Network Appliance. Commenting on a popular storage and data management blog, Riley condemned SRM as software that usually becomes "shelfware" within 90 days.

"It doesn't do anything. Each week the SRM software spits out some impressive glossies that essentially tell you how ugly you are," Riley says. "However true that may be, you can only stand so many consecutive weeks of being told you're ugly without being allowed to do anything about it." He advises those considering SRM to consider going with products that take action based on findings.

That message resonates with Randy Chalfant, a CTO at Sun Microsystems, who suggests the purpose of SRM is to groom disks. Hardware vendors, Chalfant says, often don't want IT groups to know how arrays are being partitioned because vendors tend to reserve, but do not allocate, fairly large percentages of array capacity for their own software--now and in the future. (Go to nwc.com/strategicinfo to read Chalfant's take on the value of SRM.)

His point is confirmed by a European national bank IT manager who discovered he was out of space on a new array, despite his best efforts to track utilization.

"I went to allocate a terabyte of disk to a database administrator, only to find that I didn't have it to give," he says. When he asked the vendor, EMC, about it, the company told him he didn't understand the difference between T bits--technical bits, the formatted capacity of the array--and B bits--the portion of the formatted array actually available for use by the business, excluding the capacity reserved but not yet allocated by the vendor for use in hosting the vendor's own software. The manager complained he had no way of knowing how his array was apportioned by the vendor, and was told to consult his contract, which gave permission to the vendor to reserve disk for its own use. Tools for seeing the actual disposition of disk, he was informed, are available only to the vendor's own systems engineers.

Most SRM products can't see the actual capacity allocation of arrays; they can only see what the hardware vendors let them see, says Chalfant, who adds that better SRM products have sometimes included unsanctioned workarounds that reverse-engineer vendor APIs to collect desired information. Tricks notwithstanding, he adds, people often fail to realize the power of a good SRM tool.

"Even after you exclude the sizing and partitioning done by the vendor at the frame level, we find that most customers are wasting up to 70 percent of the capacity that they do have available," he says.

The pie charts generated by an SRM package can help you figure out how much actual capacity you own and in turn, help defer purchases of new and expensive disk arrays. Given the cost of storage and the fact that it represents anywhere from 33 to 75 cents of every dollar spent on IT hardware, SRM can make a meaningful contribution to cost containment. Moreover, SRM's status-monitoring capabilities can give early warning of impending disk space emergencies or equipment faults that can grind business to a halt. (To read about how storage management can help with predictive modeling, go to nwc.com/strategicinfo.)

SRM can also assist with process improvement by making fewer persons capable of managing more storage capacity. To resonate with the front office, a full business value case must be made in terms and language that the CFO understands. SRM vendors have missed out on explaining their value in these terms--another potential explanation for the thin SRM reporting.

In the final analysis, SRM does not provide the means to solve all the problems of storage, despite the tendency of some developers to frame their product value propositions in terms of a universal plan for data domination, to paraphrase Riley. Ultimately, the information from effective SRM should be wedded to wares aimed at data or information management.

For now, the continuing mission of the storage consumer is to seek out those products that will enable them to boldly free up storage space across the far-flung storage infrastructure. Significant improvements in capacity allocation efficiency can be realized today using properly applied SRM technology. As for capacity utilization efficiency, we must continue to watch the skies.

How SRM Can Help

Storage resource management (SRM) advocates are quick to mention that the technology can chase as much as 40 percent out of the cost of storage infrastructure. Some of this value comes in the form of reduced labor requirements for administering infrastructure over time. Additional value may derive from the ability to prevent downtime by responding proactively to equipment or capacity shortfalls.

However, based on the results of storage assessments performed by Sun Microsystems' Business Analytics consulting team, CTO Randy Chalfant argues that the truly significant value of SRM is the reclamation of usable disk capacity.

Chalfant provides the following breakdown of how expensive Tier 1 (dual-port Fibre Channel or parallel SCSI) disk is being (mis)used in most companies today. As shown in this pie chart, 10 percent of a disk is typically orphaned, "often because it has been allocated to someone, but that administrator has never placed a file system on it." Additionally, 40 percent of the disk is used to store "inert data that is never referenced." He adds that as much as 15 percent has typically been allocated to a database administrator or department and simply isn't being used. And about 5 percent of capacity holds data that could be classified as "inappropriate use," such as online pictures collected by some functionary who believes his next wife's last name is JPEG.

If 60 percent of your disk capacity can be returned to productivity just by using SRM to watch capacity allocation, why aren't SRM sales skyrocketing? Perhaps consumers are being lulled into complacency by vendors' "one-stop shop" hardware marketing, or the blissful future of globally managed storage painted by the Storage Networking Industry Association's hype around the SMI-S (Storage Management Initiative).

Predicting The Future

Effective storage management may help predictive modeling. One of the leaders in this space, newcomer Onaro's SANscreen was introduced in 2005 to rave reviews. Among its features is the ability to provide a predictive analysis of how soon capacity limits will be reached based on current utilization metrics.

With trend charts that bear a close resemblance to hurricane tracking plots released by the National Weather Service and NOAA during severe weather season, Onaro claim that they provide the means to guide decision-making about new storage purchases.

Critics may question the value of past storage utilization experience as a predictor of future growth. Also, there is the question of whether the right information is being gathered by the underlying collectors to begin with, constrained as most SRM tools are by "features" designed into many vendors' arrays--intentionally put there to blur an accurate assessment of capacity utilization.

Still, whether Onaro succeeds or fails is not entirely the point. Onaro-like functionality will probably become a mainstream component of other SRM suites. Not only can predictive analysis help make SRM data more actionable, it can also be applied to other tasks.

Imagine being able to plot the savings that a vendor claims his hardware platform will accrue and then being able to compare this plot line to the actual performance of the solution. Such a capability might make it possible to hold vendors to their claims, establishing service level guarantees to which the vendor's solution must adhere-- or your money back!

Who said SRM was just pie in the sky?

Jon William Toigo is a CEO of storage consultancy Toigo Partners International, founder and chairman of the Data Management Institute, and author of 13 books. Write to him at jtoigo@intnet.net.