What elephant?

Storage is already as big as an elephant and getting bigger. It’s time to stop ignoring it and start strategising how you’re going to deal with it.

Time was, a company’s storage strategy consisted of writing a yearly cheque to EMC, Hitachi or IBM and taking delivery of storage hardware that then got shackled to the corporate servers. It was simple and predictable.

But that was before companies faced the need to double their storage capacity every six to 12 months. That was before a more heightened sense of security cast the need for reliable data backup into high relief, and before companies realised that they needed to interconnect isolated pools of storage in order to maximise ROI from expensive ERP, CRM and e-business systems.

Once an oxymoron, storage strategy is starting to get the attention it deserves, driven in part by cost. US CIO’s exclusive survey "Managing Storage" (conducted in February 2002) found that on average 22 per cent of a company’s total IT budget will be allocated to storage this year (some analysts estimate that the budget bite can go as high as 50 per cent). More than half of the survey respondents said that senior management is paying more attention to storage now. With that kind of scrutiny, smart CIOs will have to figure out how to cheaply and effectively manage their portion of what Gartner analyst Adam Couture predicts will be a 5-million terabyte elephant by 2005.

The good news is that the exponential growth in data has led to a corresponding surge in technologies and tactics that will eventually help CIOs manage storage across the entire company. The bad news is that that array of choices has transformed storage management into an increasingly complex topic. So this is a good time to take stock of where you are - and begin developing the strategies that can help you get to where you want to be in the future.

Who Brought That Elephant in Here

"[Storage strategy] is just starting to hit CIOs’ radar screens," says Mark Shirman, CEO and president of Glasshouse Technologies, a Massachusetts-based storage consultancy. "We’ve probably talked to 50 or so CIOs in the past six months, and we get one of two reactions. One is: ‘Yes, storage is a huge cost, and we’re just starting to take a look at how to better manage it.’ The other one is, ‘Huh?’" Harry Roberts falls into the first camp. "We’re taking the first steps in managing storage as an entity - we’re just putting together a five-year plan for storage, for example," says Roberts, vice president and CIO of Boscov’s Department Stores, a $US1 billion retailer headquartered in Pennsylvania. Roberts, who predicts that his storage needs will grow by 50 per cent every year, has re-examined his disaster recovery plan and is evaluating technologies that will help him manage storage growth without having to expand his staff. "We’re talking terabytes of data, and that’s a huge amount of stuff to take care of. So we really have to start thinking about it," he says.

Applications such as CRM and ERP create huge amounts of data. Moreover, the data generated by these enterprise applications is likely to be in demand all over the company so that marketing, sales and service reps can respond to customer buying habits and sales trends.

Trouble is, many data caches reside on direct attached storage (DAS), which can’t share data - or capacity - between different servers, let alone across a company. Not only does DAS make it difficult to satisfy user demands, but IT executives are forced to buy more storage - and hire more people to administer the system - even though there is still unfilled capacity in their data centre. "[That] costs a lot of money," says George Medairy, director of corporate IT at Sheetz, a $US2 billion convenience store company based in Pennsylvania. "We have hundreds of thousands of dollars invested in various aspects of storage."

Enter the storage network, which evolved in answer to the DAS problem. By pulling storage devices onto a network, companies can theoretically get rid of the interconnectivity problem and manage capacity better. Jerry McElhatton, president of global technology and operations at New York-based MasterCard International, has put together a centralised storage centre with several storage area networks that run 130 terabytes of storage. He says that even though the company’s applications have grown hugely, "we’re able to add more storage without adding more people to maintain it. The cost per byte of storage is actually less than before we put this in."

McElhatton’s solution makes a lot of sense. But storage networks - be they storage area networks (SANs), network attached storage (NAS) or some derivation of Internet protocol storage - add wrinkles of their own to the challenge of managing storage. For example, a DAS scheme generally means dealing with one or two vendors. In contrast, Glasshouse’s Shirman says, the average SAN, which requires software and networking technology as well as the storage hardware, can use products from five to seven vendors.

As a result, IS departments need more software and services in order to manage the networks, allocate storage capacity and get various vendors’ products to work together. In fact, some analysts are saying that the budget that used to go to hardware is merely moving over to accommodate the increased need for services and software. Storage hardware vendors such as EMC are reading the tea leaves the same way and branching out into software and services in an effort to retain market share.

Still, for most companies it’s not a question of whether they’ll move to networked storage, but when. "You have to go there," says Sheetz’s Medairy about the network, "because users want to get at that data. But it really changes how you manage storage." For one thing, it’s much more difficult to maintain a storage network without disrupting the users. "Now that we have more applications throughout more business units storing data on different servers on the network, timing becomes harder to deal with," he says. "It’s not a matter of just kicking a few users off the system any more. This affects lots more people."

A Trunkload of Options on the Horizon

For CIOs who are just beginning to evaluate storage with an eye to drawing up a unified plan for the future, some tantalising new developments are on the horizon. However, they come with a couple of caveats. Not all the tools are in place yet. And standards squabbles between hardware vendors are stalling important interconnectivity projects. Here’s the thing: users want them. Vendors don’t.

"We’d like to be able to mix and match hardware vendors’ solutions," says Dwight Tart, director of technical services for International Paper at its Memphis operational headquarters. However, he isn’t holding his breath. Even though Common Information Model (CIM), a storage management standard, is being developed by the Storage Networking Industry Association, a standards-making body, standards have become a struggle for supremacy by vendors’ technology strategies. Whoever controls the interface, rules. In the meantime, there are three technologies to keep an eye on.

Storage resource management (SRM) tools, currently available in early rollouts, keep a bird’s-eye view of storage capacity on the network and allocate more capacity as needed. If the SRM software sees that more storage is needed by the CRM applications, it will locate and reallocate some unused capacity.

Storage network management (SNM) tools create a map of all the devices on the storage network and monitor for errors, such as network or server failure, automating a manual process.

Storage virtualisation software acts as the Adobe Postscript for storage - it fools a variety of proprietary devices into thinking alike, thus vastly increasing interoperability among storage devices. Enterprise storage management will come into its own when these three technologies can be integrated, says Steve Duplessie, founder of Enterprise Storage Group, a Massachusetts-based research company.

And therein lies the rub. The technologies are in various stages of development, and CIOs can’t count on implementing all of them at once. Brian Lock, MasterCard International’s vice president of architecture development, thinks that SRM technology is going to be robust enough for MasterCard to implement it in the next year or so, whereas the storage virtualisation technology, while intriguing, is not quite ready for prime time. "Getting a tool for storage virtualisation is a little further out. The companies are newer, and it’s a tricky technology," he says.

Tart has recently started using Computer Associates’ SRM software in an effort to better manage storage capacity on his SANs. He says that although he’s "very early" in implementation, his staff has already identified 3 terabytes of unused storage hiding out in the company’s 50-odd terabytes of storage. "We’re going to have to continually better manage our networked capacity, and that’s going to require more than just manpower," he says. "We’ll need automated tools to get it done."

The (Future) Lord of the Storage Jungle

How will the CIO know when the storage jungle has been mastered? Duplessie thinks the future will look a lot like this: IT will be able to analyse storage usage at a granular level, accounting for storage needs by department or line of business.

"Today, we can’t delineate between storing Steve the Moron’s goofy .mpeg download and an Oracle database transaction," says Duplessie.

In addition, IS departments will become storage service providers. As companies begin to stratify data by quality and apportion storage accordingly, CIOs will be able to work out quality and service agreements with each department. That will allow the CIO to attach a monetary value to storage services and identify storage transgressors.

The big goal for companies is to formulate and automate storage policies. According to Duplessie, policy is a hot term these days. The idea is to create a corporate template of policies and procedures about storage and commit them to a rules engine that will then monitor storage and enforce the rules. Such rules can stratify different levels of users and assign certain levels of access to storage, as well as backup schedules and data retention periods. Storage policies will also help get rid of data that’s just wasting space, says Duplessie. They’ll even weed out the redundant data storage practices and make them, one might say, tomorrow’s white elephant.

The Sharing Environment

The basics of storage networks.

NETWORK ATTACHED STORAGE (NAS) generally refers to storage devices that can be attached to the existing corporate network, making NAS pretty cheap to implement. But corporate networks usually run according to Internet protocol, which sends data willy-nilly or in a structure known as file format. But storage data is generally sent in block format, which requires that the data be sent in a certain order. Storage data does not generally take kindly to being sent in file format. And sending storage data over the company network can seriously slow network speeds.

INTERNET PROTOCOL (IP) storage is a transport mechanism that seeks to solve the aforementioned problem of sending block data over a regular network. Using IP technology to transfer storage data in the block format it prefers, IP storage can save money by allowing a company to use its existing network infrastructure for storage - but it also raises the same performance issues as NAS (for more on IP storage, see "Storage Gets Caught in the Net", CIO May).

STORAGE AREA NETWORK (SAN) bypasses sending data over regular networks by building a separate network just for storage devices, servers, backup systems and so forth, and then connecting the SAN to the regular network. SANs may be based on an industrial-strength connector such as Fibre Channel, which can handle the heavy bandwidth demands of storage data. Keeping storage data off regular corporate networks does lighten the traffic load, but SANs are both expensive and complicated to construct.

STORAGE VIRTUALISATION SOFTWARE cajoles different storage devices to interoperate despite the fact that the various operating systems manage storage differently, are set up according to each maker’s technical protocol, and cannot communicate with each other. Storage virtualisation schemes create a separate layer of software that acts as a universal translator. For example, you can have a pool of storage from Compaq and a pool of storage from EMC, and you’ll be able to move data back and forth between the two.

Along with the rise in the need for storage goes an increase in storage spending.

What percentage of your IT budget goes to storage?

• 17% in 2001
• 22% in 2002

The Rise of Network Storage

Enterprise applications generate huge amounts of data that need extra storageWhat’s driving your increased demand for storage?

31% Internet applications 32% CRM 33% ERP 44% Business process backup

Although DAS remains popular, close to half of respondents listed SANs as the next storage solution they’d implement.

What’s your primary storage technology?

14% Network attached storage or NAS 24% Storage area networks or SANs 53% Direct attached storage or DAS

Respondents: 100 CIOs and IT executives. The "Managing Storage" exclusive CIO (US) survey was administered online during February 2002. For the full survey, go to www2.cio.com/research.

3 Ways to Take Control of Storage and why you should do each of them.

1: INVENTORY AND CLASSIFY DATE - You never know what you can get rid of. Jerry McElhatton, president of global technology and operations at MasterCard International, was able to unearth and remove redundant pieces of data, thus freeing up storage capacity. "When you grow storage on an application rather than companywide basis, you end up with databases with a lot of the same content in them," says McElhatton. A data inventory will flush out the redundant content, and "then you can put it into a single file and let that one data component serve all the people who need it", he adds.

2: TRY NEW MANAGEMENT TOOLS- You'll never know what you'll find. George Medairy, director of corporate IT at Sheetz, a convenience store company, discovered a Computer Associates product called Brightstor "bundled in with stuff we already had". He’s found it very useful in automating certain data management tasks such as monitoring storage tapes. "It’s really helped us, and we didn’t have to go looking for a product to do this job," he says.

3: EVALUATE DATA BACKUP AND BUSINESS CONTINUITY NEEDS - You’ll never know what you’ll discover.

Bill Gearhart, director of IT at Rinker Materials, a Florida-based construction materials manufacturer with annual revenue of more than $US2 billion, recently redid his disaster recovery plan and discovered that he was neglecting some important data. Bringing e-mail up wasn’t a huge priority with the old plan. But since many people need the system for communication - and for storing business-critical information - Gearhart says that e-mail recovery has vaulted to the top of his company’s disaster recovery list.