An Overview Of Network Storage Options For The Acronym-Impaired.

As businesses continue to collect reams of data about customers, about clients, about projects, and all the information that is associated with the billions of transactions that occur every day, they are facing increasing challenges related to storing and accessing that data. It is no longer unusual for companies to have terabytes of stored data. What does this really mean? There is a rule of thumb that estimates 50,000 bytes of storage are required for one scanned letter size page and 10,000 pages can be stored in one four-drawer file cabinet. Based on these estimates, a gigabyte (1,000MB) is required to store the scanned contents of two file cabinets. A terabyte of data, therefore, is the paper equivalent of 2,000 four-drawer file cabinets.

Now, imagine that your company needs continuous access to these file cabinets, retrieving data and reftiling it after use, and, at the same time, you need to continue to be productive at your business. You can't tell your customers to wait for a day or two while you go through a couple of thousand file cabinets before you get back to them. Also, you would really like to have an extra copy of everything in your files for backup, but you can't afford to have a person stand at the copy machine for the next few months. This is analogous to the problem that is encountered by the main servers that support office Local Area Networks or LANs. Main servers are continuously bombarded with demands: run this application, get me this data, file this for me, go to this website, open my e-mail, and download that document. As the demand for data access increases, less of the main server time is available for running applications and productivity and storage requirements go head to head in a battle for the server's limited re sources.

Levels 1 And 2--Bus-Attached Storage And Network Attached Storage

Depending on how much data your company has to deal with, there are three types of storage options available (See Fig). The simplest level is a "bus attachment" where your storage device is directly connected to your main server. Bus-attached storage operates through the main server and its availability and performance are limited to the server's capabilities and loading.

The next step up is to remove some of your stored data (typically archived files, completed projects, or backup) and place it on a separate file server attached not to your main server, but to your network. This Network Attached Storage (NAS) arrangement frees your main server from data storage chores such as sending, receiving, and storing data files, although some main server attention is still required to direct requests to the NAS file server.

There are many NAS devices available with varying capacities and expandability. A NAS server is basically a cabinet of hard disks equipped with a processor and a special operating system, which can communicate with the operating system of the main server. Some NAS servers can support multiple operating systems, which is useful for organizations that have data files that must be shared by Windows and Unix users. It is particularly attractive as a quick fix when the main network server is running out of storage space and there isn't any down time available for maintenance because a NAS device is easy to install.

One Step Beyond

Beyond NAS, the need for maximum data storage and high-speed performance requires the development of separate networks dedicated only to storage. The driving force behind the push for Storage Area Networks is the need to keep networks online and functioning at full speed while, at the same time, offering multiple users continuous access to stored data and ensuring that data is protected from loss. By the way, the term "Storage Area Network" is somewhat contrived in an effort to follow the precedent set by other spatially-defined network terminology such as Local Area Network and Wide Area Network. It would have been better and more simply named a "storage network."

A SAN is worth considering if your work commonly involves sharing a large amount of data requiring fast transfers between multiple systems or workstations. Data-intensive applications that might be good candidates for employing a SAN include e-commerce ventures with large customer databases, banking and financial institutions, geographic information, and Global Positioning Systems (018 and GPS) businesses, medical and healthcare organizations, or enterprises that use biometric data such as fingerprints, iris and retina scanning, voice recognition, or handwriting analysis. SANs are useful for supplying access to critical data to anyone in the enterprise at any time no matter where they're located or what operating system they happen to be using.

A SAN, then, is a system for interconnecting and allowing different kinds of storage devices to be shared by all users by way of a storage server or servers. (To further complicate the jargon, the disk or tape devices that are connected to the SAN may also be referred to as SAN-Attached Storage, or SAS. This is even more confusing, considering that some people take SAS to mean "Server Attached Storage," which is really bus-attached storage.) With a SAN, when users need to access data, they bypass the main server and contact the storage server, which puts them in touch with the device that contains the required data. In addition, because this arrangement also allows the storage devices to communicate with each other, it is ideal for performing time-consuming functions, like backing up and restoring data or transferring data from one storage device to another. This leaves the main server free to answer demands for applications and communications. Ideally, SANs give organizations the ability to scale their stora ge facilities to meet growing needs to store more data. SANs can even be designed to incorporate NAS subnetworks, that is, NAS file servers can be attached to the SAN network for a more complex storage hierarchy.

Connecting The Devices--A Vital Component

The SAN is not really a new idea, but is becoming more mainstream now because of Fibre Channel, a network technology of high-speed connections. Bus-attached storage devices typically use SCSI. SCSI copper cables can only be effective up to a distance of 25 meters (and the effective distance for many cable types is actually quite less), which severely limits how far away from the server a peripheral device can be physically located. In comparison, Fibre Channel can extend up to ten kilometers. As a rule, Fibre Channel data transmission is about three times faster than SCSI cables, although so-called Ultra SCSI cables are also capable of high speed. In general, however, Fibre Channel has fewer length and device restrictions than SCSI and is the technology of choice for setting up storage networks.

Optimizing SAN Performance--Fast And Accurate Testing Is Essential

Currently, some limitations have prevented optimal SAN performance. SANs from different vendors don't interoperate easily and SAN vendors are grappling with establishing Fibre Channel standards and resolving compatibility problems. Users, so far, have primarily had to stick with a single vendor's solution because SANs are not yet universally interoperable. However, some vendors are now stepping into the role of middlemen, putting together packages that are designed to take the guesswork out of choosing what pieces are needed for a particular storage application.

In spite of these advances, a significant challenge remains for SAN optimization, especially for systems made up of components from different vendors and that is the demand for testing, monitoring and analyzing performance. In fact, testing and analytical tools are often a critical link in the implementation of advanced storage systems. Specific test protocols are required to address a wide variety of testing and analysis scenarios. Manufacturers must perform design verification testing and production line testing, which includes functional, performance, stress, and data integrity testing. Systems integrators need to assure protocol compliance and error recovery and must also be able to integrate Fibre Channel with SCSI network infrastructures. End users also must conduct performance testing and product compliance with the added complications of field service troubleshooting and fault isolation.

To date, the diagnostic tools and techniques for the dynamic testing of advanced data storage systems has not kept up with the technological advance of the data storage systems themselves. Routine problem troubleshooting has been limited to a narrow range of non-compatible, difficult to use software and hardware products from a multitude of vendors. These traditional means of data storage system testing suffer from numerous limitations, including:

* Existing test hardware and software products are complicated and require a high level of technical expertise to operate and understand.

* Most test applications are not automated and necessitate custom applications to be developed for each test case.

* Current methods of test application development are time consuming and expensive.

* Tests are not portable across applications, interfaces, processor families, product configurations, and operating system environments.

Fortunately, the technology behind the testing and analysis of data storage systems is beginning to catch up with the demand. New approaches to data storage testing, monitoring, and analysis can now be found, which directly address the shortcomings and limitations of the current methods and products in the marketplace today, making the future benefits of SANs decidedly closer to being realized.

Rick Brechtlein is the president and CEO of Shugart Technology