0 Technology Publications

Trying to predict where technology, in general, will be in a number of years is probably more art than science. Whether it's consumer-or business-related, technology for the sake of technology is useless unless it addresses a real world problem or fills a specific need.

However, predictions are inevitable, so we decided to give it a go. There are two ways to approach predictions. The first is to use a strictly linear progression (i.e., in five years existing technologies will have advanced at the same rate as the preceding five years), or we can take the "where would we like to be" approach (i.e., in five years we should have 100Gb wireless Ethernet!). We also have to consider that advances in one technology generally depend on, and are influenced by, advances in connected technologies (what good is a 100Gb wireless Ethernet if the adapter is larger than the new wrist-watch-sized PDA, cell phone, or laptop device currently in use?). Finally, there is always the likelihood that a new, innovative technology will be introduced that has no precedent at all.

[ILLUSTRATION OMITTED]

Demand Just Gets Bigger

The one thing we can all agree on is that the demand for storage will simply continue to grow. By 2010, individual users requiring multiple terabytes of storage will be commonplace. Small to medium-size companies will have petabytes of data; large enterprises will have exabytes. What's after that ... zettabytes!?

Advances in disk hardware will address a portion of this increase. There will be multi-terabyte disks by 2010--possibly even petabyte disks. This would, by extension, lead to petabyte- and exabyte-scale RAID systems. Of course, all of this will need to be managed in a new way--old models will simply be swamped at this scale.

Storage Management Shifts

Storage management has already shifted away from individual application servers; storage area networks (SANs) have allowed administrators to move these functions. Given the number of application servers in even a medium-size company, it is not realistic to require that each application server manage its own storage. The application server (database, e-mail, web or other) should be used to perform its specific function, not general maintenance.

In SANs, storage management is also shifting away from individual storage systems. Similar to application servers, these RAID systems would become islands within the SAN and require administrators to manage them individually.

[ILLUSTRATION OMITTED]

These functions are now moving into the SAN itself, either through intelligent switches or appliances designed to provide a common layer of storage services across various application servers and storage systems within the SAN. By 2010, the move will be complete and storage management will use a distributed model wherein the SAN provides the storage services for the application servers.

Storage Subscriptions

Part of the value of moving storage management to the SAN, and away from the application servers, is that it relieves the application server from managing all aspects of its storage. But it's not enough to just move the disk out of the application server. The first step removed the need for the application server to manage mirrors and RAID levels for its LUNs. The next steps remove the need for the application server to be involved in backup and disaster recovery.

By 2010, changes in operating systems and server-based applications, coupled with standard interfaces for managing storage, will allow the application server to use storage in much the same way our homes use electricity. The application servers will be authorized to use a certain amount of storage from the SAN. If the server needs more storage, it will negotiate with the SAN for a temporary or short-term increase. If the server is not fully using its allotment, the SAN may negotiate to regain some of the excess. In either case, the administrator will not be required to make any changes to the environment. In an ideal world, all this will take place automatically and seamlessly, with no adverse impact to service.

Storage by Policy

Clearly it will not be feasible for administrators to manage these SANs as discrete elements. In the future, storage systems will use policies to determine exactly how storage resources should be used, how to identify which application servers have higher priority, and what level of protection each system needs. The storage system will automatically move data, route traffic and rebalance resources as needed to meet the objectives and priorities outlined in these policies. Today's basic multi-pathing and load-balancing model will transform into a dynamic, coordinated system that adapts not only to traffic changes, but delivers flexible QoS, while allowing administrators to specify when the needs of different systems or even whole departments change.

Backup, Backup, and More Backup!

Ten years ago, the biggest challenge in storage management was backup. Today, the biggest challenge in storage management is ... backup. Given the trend, there is no reason to believe that this will change by 2010. What will change is the model used to address the backup challenge.