Are You Managing Your Storage Resources? You Can No Longer Afford Not To.

There are only two relevant questions requiring affirmative answers to justify a storage resource management (SRM) solution. Did it save money and did it improve application performance? Historically companies made the mistake of focusing only on the first of these criteria, namely saving hardware dollars. Since information technology (IT) was a large cost center, budget pressures often led financial managers to look to IT whenever spending constraints were required. This created a tug-of-war between user demands for improved system availability, which always seems to cost more, and financial mandates. In addition, budget constraints have affected the staff available to manage increasingly large and complex storage environments. Overtime was preferable to additional staff or software investments. Management software that reduced the requirement for "carrying a beeper" after hours was not viewed as "substantive" savings.

Things have changed. Many of today's most successful companies have developed a new view of IT. In fact, the "new economy" is based on a strategy of exploiting IT's revenue-generating potential to create new sales channels and expand sales opportunities, as well as streamline the supply chain and other operations that impact profitability. This has created the opposite pressure on IT: guarantee system availability and optimum end-user response times regardless of costs. Eventually, so the thinking goes, revenue growth will justify the spending. We are now learning that this extreme is no more viable than its penny-pinching predecessor.

Decisions, Decisions: A Hardware Or Software Solution?

Regardless of whether the pressures are to meet a budget or a response time objective, systems and storage professionals are presented with two basic ways to address their availability needs: hardware or software.

Limited staff size and expertise foster the hardware approach, and hardware vendors sell to the executive, suite where performance and service guarantees can be effectively positioned as expensive but justified by "turnkey" installations. It would be difficult for management not to be attracted to solutions that avoid requirements for staff development and problems of staff retention by deploying a hardware solution especially when hardware costs are at all time lows and vendors can provide a service package.

Specifically in the storage arena, disk hardware prices have declined dramatically over the last 15 years. Premium RAID devices have fallen from $70 to $80 for a fully configured and installed megabyte (MB) in 1975 to well under $l today. It's no surprise that adding more hardware has been the primary solution, but throwing hardware at these problems has failed to solve them. Eventually this ever-expanding storage capacity creates more problems than it solves. Some analysts have described it as a tsunami; others, an avalanche. The point is that this strategy will bury you in disks without assuring that your critical applications are available and performing as required even with a constant infusion of more and more hardware.

Storage Resource Management Software Can Reign In Runaway Capacity

Let's consider an illustration of the problem beginning with a simple premise: the reason you need more storage is that you have more data. Once you create a new piece of data, you have to protect it and make it available if the primary location or copy is unavailable, or if you need to recover from a site-wide disaster. You may have a traditional backup strategy that requires that full and incremental backups occur on some fixed schedule in addition to these remote copies, mirrors and other "hot spares." So, for each MB of data you create, you could have 1MB mirrored locally, 1MB remotely, 1MB backed up when created and (at least) 1MB backed up whenever there is a change. So from its inception, a hypothetical 1MB file could become 4MB of data as a function of data protection and availability strategies.

As a rule, however, you cannot manage at the file level without storage resource management (SRM) software. Hardware mirroring is usually instrumented for an entire volume, and system management tools are generally at the file system level. So, without SRM, regardless of the value of the individual file, you may be creating multiple copies of files that aren't particularly important in order to ensure that the critical files are protected and available. As a result, hardware-only strategies and basic systems management without SRM result in more and more old, obsolete and duplicate files.

Basic file-level management hasn't been a distributed systems discipline except for the individual user trying to find the file with the "forgotten name." Think about your own PC or laptop. With a few gigabytes and 50,000 files, which isn't an unusually large number for a typical business workstation, how often do you clean out old, obsolete or duplicate files? And for disk arrays and super servers, the problem is the same only on a much larger scale. As a result, they not only stay on the disk, but you add more of them. For many of these, you also are creating mirrors and backing them up. Look at the server where you store your three-year-old e-mails and five-year-old presentations to get a flavor of how much data is out there consuming space.

When you bring in more capacity, it is quickly filled with new data plus the copies required by your data availability strategy. Why this happens is a mystery to many, but it is a pervasive and immutable truth: the more storage you have, the more you will need. Or to borrow a quote from a storage administrator, "users will expand their requirement for storage to fill whatever capacity is available." That's as true for an enterprise server as it is for a home PC. After all what are the two main questions we always ask about a new system: how fast is it and how large is the hard drive?

So how do we break the cycle of "just buy more" storage to achieve cost control and availability?

Storage area networks (SANs) are supposed to be the answer. We are told to just implement a SAN or network attached storage (NAS) solution and we will have "virtually unlimited storage." IT managers have tried this once already in the mainframe world, and for some of the same reasons: consolidating storage so that more capacity is effectively shared among systems.

With the capacity glut of a SAN, it may take users a bit longer to fill it, but fill it they will. Just as mainframe storage resource management became even more critical to ensure availability and performance when DASD capacity increased and became a shared commodity, the need for this management discipline will become increasing critical for networked storage, not eliminated by it.

Whether you are talking about sizing partitions in shared devices, identifying growing tables in a database file system, or managing SAN zones, an understanding of how storage is being used at the file level by users, servers and applications will become more and more important for managing availability. Old, obsolete and duplicate files will create more problems not only because of storage use but also because of the system resources that are required to back them up, including the actual backup processing itself and the additional disk capacity to store the backups. True, SANs will address the shrinking backup window by offloading this process from the business network and implementing server-less strategies using storage devices that have the intelligence. Online applications will be protected from performance problems caused by backup processing, but will they be protected from each other? When multiple paths are available, how will you know which processes get higher priorities when there is contention, or fail-overs?

Over time, even with the considerable dedicated bandwidth and processing power of a SAN and the benefits of server-less backup, the "pipe" still has a finite capacity and disk storage remains a physical entity that will fill up. If data continues to proliferate, which assuredly it will, the system resources and time required to complete it will continue to increase until the SAN will no longer support it. And then, what do you do, continue to expand your SAN, add more servers, routers, hubs, bridges, disk arrays, tape robots? Where does it all end? When you can no longer afford to build more buildings to keep it in or hit the distance limitations on the wire? Just how much complexity can you manage without management tools?

So what is the answer?

* Determine through automated monitoring and discovery what capacity you have, how it is deployed, who/what is using it, and how fast that use is growing.

* Automatically identify wasted capacity and recover it by archiving or deleting old, obsolete and duplicate data as appropriate.

* Map enterprise-wide storage use and 110 performance to the applications that run your business.

* Establish policies to automatically manage resource use and performance according to an application's priority.

By taking these steps, you create an application-centric storage management strategy that spans all your storage resources. Using an application-centric approach, you can bring a wide range of data from a diverse set of technologies into your management schema and have a practical way of executing automated management strategies based on analyses of this data. This approach will result in cost savings along with assured application availability by centralizing management of all storage-related resources and organizing them according to business priorities. Here are some examples:

If you can centralize your storage management, you can integrate information, simplify tasks, and ensure that effective storage management can be achieved with fewer staff. If nothing else, a common console reduces the amount of time required to log on to different machines, launch different programs, re-key or export data for analysis. Beyond that, users can be insulated from the device-specific interfaces, commands and product-specific scripts whether for a disk array, server or interconnect. For example, changing or setting volume mirroring or remote copy status for a volume on a device can be administered consistently because the console will handle the interface to the vendor tool. Staff will spend their time more productively and have shorter learning curves.

If you know what you have and where it is, you will discover surprising sources of underutilized and forgotten capacity so that you don't always have to buy more.

Automatically discovering all the storage-related devices and their connectivity ensures that these assets are neither lost nor underutilized. Storage assets can become very difficult to track, much less manage or optimize, especially when corporations span multiple geographic locations and have diversified channels, product lines or manufacturing models. A historical view of capacity use and performance of these assets provides the information required for redeployment or consolidation. New initiatives would benefit from lower startup costs and growth spurts could be accommodated without an emergency hardware shipment.

If you know how your storage is being used you can reduce total storage costs.

Old, obsolete and duplicate files waste capacity. Servers can be consumed with temp files that no one has time to clean up, because there is no easy way to execute these tasks across the heterogeneous enterprise. Reclaiming this disk capacity must occur in an automated fashion to be practical. These savings extend beyond buying back disks. The result will also include reduced data protection and availability costs, including backup processing requirements, as well as mirrored volumes and remote copy capacity. And, when you understand an application's growth patterns across platforms and network dependencies, both daily decision making and future planning requirements benefit.

If you use an application centric management strategy, you can quickly identify, correct and even prevent problems that impact your business.

Whether a user defined threshold violation, a media failure, or a hardware device "phoning home" for service, correlation of these events with affected applications provides immediate notification of potential disruptions and can initiate automated correction or prevention. With an application perspective, automation can be policy-based with a resulting increase in availability of the applications that are most important to your business. Automating repetitive, routine tasks that account for 80% of storage management allows time to be spent on the exceptions and value-added tasks.

If you know what data is used by specific applications, you can prioritize resources to ensure that response times are meeting business requirements.

More effective decisions about device type, partition size, file system contents, SAN zoning and consolidation can be made if resource deployment matches application priorities. And, the more comprehensive this application view is, the more efficiently and proactively you can analyze trends, develop forecasts, and model the impact of one application on another.

If you have a complete map of the technology from logical volume to physical volume to the actual disk, you will continue to optimize your ROI for storage and ensure application availability for tomorrow's high capacity disks.

As storage technologies place more storage capacity in an ever-smaller footprint, a new management capability will be required. With a one-terabyte disk on the horizon, managing data, placement will become more critical. Mapping storage to an application will extend beyond the physical volume map needed today to identification and control of the placement of the data on the disk itself. Otherwise, you will be unable to take advantage of this capacity without incurring the performance problems that are likely to result from increased I/O to these super sized disks.

With an application-centric approach to storage management, you can establish a management discipline that will control costs as well as enhance the responsiveness of IT systems to emerging business opportunities. When you know how your storage resources are being used, you can make intelligent decisions about acquisitions as well as build a management infrastructure that can intelligently manage resources for optimum availability. Your storage management strategy can impact your bottom line by reducing costs and by optimizing the IT processed that make money, too.

Beverly Casstevens is senior product marketing manager, Storage Solutions, for BMC Software (Houston, TX).