Tiered storage: new strategies match new demands and opportunities.

The world of storage is changing, and end-user IT departments need help in planning new strategies to match the new demands and to take advantage of new opportunities to use storage resources more intelligently. One of the most promising new directions involves solutions that combine the characteristics of multiple storage resources in a single managed system. This approach promises to offer concrete help for a large number of different storage issues that are all too familiar to IT departments.

One of these is the fact that the demands on storage continue to increase. Data keeps growing in volume, and it's becoming more valuable to more different parts of the enterprise. The millions of files managed by the typical large organization contain customer records, documents, drawings, images, reference material, messages, etc. Used the right way, that data can give organizations competitive advantages--ways of providing their customers with better products and faster service. Part of what that means is that protecting the data, retaining it over time, and making certain that it can be accessed quickly when needed has become a critical issue for the entire IT team.

A second is that the idea of protection itself has also become more complex. People have recognized for some time that their data is in danger not only from natural disasters but also from computer viruses, human error, and malicious acts. But now they are being forced to contend with a whole new set of cost factors, as regulatory agencies demand that users add legal requirements for protection, retention, and recovery of data to the existing set of business requirements.

And finally, the kind of economic pressure IT organizations face is changing. Companies are recognizing that, over time, the cost of capacity itself is much less important than the cost of managing their systems, and organizations are placing a premium on solutions that can, in the long run, free the IT staff from lower-level management jobs and allow them to add higher levels of value.

The Multi-Resource Storage Environment

A key concept that is helping IT departments deal with the complexity of managing data in the face of these pressures is the idea of combining different kinds of storage resources in a single managed system. The basic idea is simple: Use different resources to provide different kinds of access, retention, cost and protection characteristics and match the resources to the right kind of data. The goal is to be able to make sure that the data can serve all of its various uses in the most efficient way, while making the most efficient use of physical and human resources.

[FIGURE 1 OMITTED]

Behind that simple-sounding idea, however, is a complex set of management tasks that requires recognizing different kinds of storage resources, creating a storage environment that combines them, automating the movement between them, and providing access to the data from the different locations. In the past, the term "tiered storage" has sometimes been applied to multiple resource use, but that term is really something of a misnomer since it suggests a fixed hierarchy of storage locations through which data is always passed. The newer multi-resource systems operate very differently. In them, the various storage resources tend to sit side by side, and access can be provided directly to any of them without necessarily moving the data back through different levels. Instead of tiers of storage, there are multiple alternative resources which are used to store and access data based on a changing set of needs.

The kind of managed storage environment can look very different depending on what kind of uses and applications are being served. At one end of the spectrum is the kind of fully unified Data Lifecycle Management (DLM) environment that ADIC's StorNext Management Suite (SNMS) creates. This unified storage environment can consist of multiple kinds of disk systems (performance disk and value disk) as well as tape (both libraries and off-site tape pools). The unified environment is created by file system technology that allows all the files stored to be accessed directly from any location by the applications that have access to the storage environment. Management of data on different resources--migration, copying, movement off site--is automatically managed by SNMS based on policies that allow the access patterns and age of files to map resource use and file management to business value for the data. Policies can be set to include protection of data on different media types as an automated attribute of the managed system.

This data environment provides general-purpose storage and it presents itself as disk to applications using the storage. The hosts look at this pool of storage and see usable capacity--behind it are different disk and tape resources with different access and protection characteristics and policies to move data between them. Because it is based on a fully heterogeneous file system, the storage environment created by SNMS supports multiple server platforms and operating systems, allowing applications designed for UNIX, Linux and Windows systems to share a common pool of automatically managed data. There are several uses for this kind of environment. Specialists have used this kind of consolidated storage pool for managing very large data sets and for sharing data in highly collaborative work environments. IT is now beginning to reach a more general IT audience, as new demands for data management make consolidated storage attractive and as content-aware Information Lifecycle Management (ILM) applications need a foundation to actually map data to different resources.

Integrated Disk-Based Backup Solutions

Another very practical use of multi-resource storage environments is the creation of specialized solutions that present themselves to applications as a single resource, but that have a range of characteristics that no single storage resource can provide. A perfect example is the idea of including disk as part of the backup process.

With the advent of lower-cost ATA disk systems, the idea of including disk within the backup process has become more appealing. Disk offers random access to data blocks, allowing it to handle some kinds of write and read operations much more efficiently than tape. And it can be organized easily into highly resilient redundant arrays so that data can remain readily available even through a fault condition that affects a single spindle.

But, for a variety of reasons, backup still needs tape for the vast majority of end-user applications and environments. Tape drives handle large data streams very effectively, and backup applications have mature tape library interfaces. Tape is a static medium that is unaffected by viruses, and it provides the lowest cost bulk storage available--much less expensive than even the lowest cost disk. And tape is a wonderful medium for long-term storage: durable, stable, easy to transport and store.

Various systems have been offered to give backup the characteristics of both media types, but most of them have created new problems and increased the integration and management burden for IT departments. What they share in common is that they write data to a single disk resource first, and then operate a separate process to move the data a second time to a tape resource. This two-stage operation has many problems. It doubles the amount of data movement that the backup servers perform, so it degrades their other operations, clogs networks, and increases the management burden on IT departments and integrators. It also may increase licensing costs for backup software, since most of the backup software that first writes to disk and later to tape requires separate licensing for the disk and tape resources. And of course it makes management of the task much more complex, leading to confusion when it comes to providing integration and support services.

[FIGURE 2 OMITTED]

Using a unified storage environment with managed data migration to combine disk and tape offers a much more elegant way to create a better backup solution. ADIC's Pathlight VX is a good example. The Pathlight VX looks like a single resource--a tape library--to backup software but, in reality, it creates an integrated disk-tape storage environment with the characteristics of both media types. When backup jobs are first written, the data is written on a RAID-5 ATA array that presents itself to the backup application as if it were a tape library with multiple drives and media. Then, in the background and off-line, the Pathlight VX system writes the data to real media in a real tape library and exports the tapes for long-term storage triggered by commands issued by the backup software. All data movement is carried out by data movers within the Pathlight VX system, so there is no negative impact on production servers, and the data movement occurs completely off-line so there is no increase in network traffic. Most importantly, the system integrates the functions so there is one management interface for all operations--the conventional backup application software that users currently use.

The end result is a single system, with a single integration and service source that combines the qualities of the two storage resources. Initial backups are written to disk and, depending on how the user sets up the system, most restores will be from disk--providing faster raw performance and RAID levels of fault tolerance. But for large-scale storage and long-term protection, tapes are created and save sets are moved off-site just as they are in any conventional backup scenario. The two resources are integrated, but they also exist side-by-side--the tapes that are created use standard backup software format and can be read directly in any compatible tape drive or they may be restored through the disk system. This combination of value, performance and streamlined management is giving IT departments faced with shrinking backup windows and increased data volumes a new source of help.

Conclusion

Storage remains one of the key issues facing IT departments, who are facing continued data growth, more challenging data protection requirements, and a mandate to reduce the total costs of system ownership by reducing management. A new approach that is helping users solve these problems integrates multiple kinds of storage resources into a single managed entity. ADIC's StorNext Management Suite data management software uses this approach to create a comprehensive Data Lifecycle Management environment that automatically manages and protects data on different resources over its lifetime. ADIC's Pathlight VX combines disk and tape into a single backup solution that gives conventional backup software and backup processes the performance and fault tolerance of disk along with the value and long-term retention characteristics of tape.

Kevin Honeycutt is vice president, Enterprise Partners. ADIC (Redmond, WA)

www.adic.com