SANS And Tape Storage New Storage Paradigm For Enterprise Data - Technology Information

Computer Technology Review, Jan, 2001 by Dave Uvelli

Today, SAN architecture is being used together with automated tape storage primarily for backup applications, but this combination will soon provide powerful new functionality to help the enterprise solve a much broader range of large-scale data storage problems. The key is found in initiatives underway now to combine Storage Area Networks with automated data management tools and low-cost, large capacity, multitiered storage systems.

These systems are integrating Fibre disk and a new generation of automated tape storage into a single system. Some will leverage new technology at both ends of the storage continuum--they will use the capacity and flexibility of Fibre disk to provide true file sharing over a switched fabric, and they will leverage the capacity, low cost, scalability, and reliability of new tape systems for storing high volume data. Best of all, they will integrate these functions together using automated data management tools.

Until fairly recently, very large scale data management has been a real problem for a relatively small number of the largest sites--primarily organizations involved in very large imaging applications, such as organizing satellite download data or making sense out of multi-gigabyte seismic data sets. But that's changing rapidly.

Today, corporate IT departments are facing multiterabyte storage problems as a matter of course. The applications creating this data run the gamut from general-purpose databases and standard office documents to more specialized uses, including data warehousing, graphics, and serving rich content to web-based servers.

But if it's just a question of capacity, why do we need a new approach? After all, disk is plentiful and cheap and it's easy to hang more NAS filers on the network. The answer is that the amount of data is outrunning the capacity of the networks and the network administrators to handle it. It just isn't an issue of having capacity to store the data--it's backing it up, protecting it, moving it, and figuring how to make it available for use.

The approach to solving the problem that makes the most sense builds on what the largest data users have been doing for some time. They recognize that all data isn't alike and they have systems for treating different categories of information in different ways depending on how they need to use it. When data from a satellite image system is being actively analyzed or presented in a graphic format, it is staged on large capacity, fast access disk. But the data sets that are not actively being used are normally housed on high capacity storage systems built around tape storage.

These larger storage systems provide lower cost storage but they also reduce active management requirements since data safely stored on removable media no longer requires high levels of active management. Moving data between the active disk resource and the high capacity storage devices may be a manual process, but in some cases IT managers use automated data migration tools to move data between the different devices based on a series of usage and file characteristic guidelines.

The change that is beginning to take place in enterprise storage is that a new generation of tools is beginning to become available that performs the same kinds of large-scale data management tasks, but that makes the process more integrated, more flexible, less expensive, and easier to administer.

The new generation of storage management products is based on three key principles:

* Not all data is the same or has the same requirements for access.

* The most active data needs the highest bandwidth and the most flexible sharing.

* Automated tools need to help people move data between different storage locations.

Let's look at each of these ideas and how they relate to this new generation of storage systems.

Not all data is equal. You could argue that the real problem with storage is that most systems treat almost all the data in the same way--as if all of it were serving the same purpose. The reality is very different. In most networks, a relatively small amount of data is actively used every day, changed frequently, or accessed by a large number of users regularly. The majority of data, though, is used periodically. Still more, usually older data, like last year's PowerPoint presentations, needs to be preserved and available, but is accessed only occasionally, maybe even only when other copies of information are lost or damaged. What enterprise IT departments are beginning to discover is that if you treat all data in the same way, infrastructure costs and management loads become oppressive as they try to keep up with the volume of static data.

The most active data needs bandwidth and shared access. For the most active data, the new tools provide the high bandwidth and data sharing of SAN technology. Up to now, SANs have been used mostly for backup or for providing limited access to a single, expensive disk. The new tools leverage Fibre disk as a more powerful resource by letting data down to the individual file level to be shared over a switched fabric among a variety of users without moving or copying files. The high SAN bandwidth means that this shared data can be accessed at local disk speed by all of the connected hosts. This capability of SAN file sharing systems has been limited up to now to high volume applications like video editing and specialized imaging applications, but it is now emerging as a powerful tool for more mainstream applications.