SAN-itize Your Network

The amount of storage capacity shipped this year has doubled over that oflast year. And a significant portion (perhaps all) of network-stored data is consideredmission-critical. In this environment, just deploying more and faster storage devices isnot enough.

One approach, network-attached storage (NAS), uses an integrated storage system thatcommunicates with network protocols like TCP/IP. A network device is connected to thestorage system and functions as a server. The device processes file I/O protocols such asNFS, thereby managing data transfers between itself and clients. The drawback? Theprotocols limit performance.

Storage-area networks (SANs) meet the challenge in a different way, with betterperformance results. A SAN is a high-speed "subnet" that establishes a directconnection between heterogeneous storage resources and servers. SAN network processorshandle data channels and network protocols and provide a focused set of services forhigh-performance, reliable shared storage access. They uniquely exploit data channelfunctionality and performance; they do not function as servers. Think of a SAN as anextended and shared storage bus.

DATA MIGRATION MATTERS
With the SAN architecture, data becomes more accessible to users anywhere on thenetwork and different kinds of storage resources can be shared between differentapplications and servers. For example, a large bank that recently acquired a smaller bankselected a SAN solution to act as the traffic cop and feed continuous live transactiondata from 1,200 branches in 15 states to headquarters.

The system also solved a major data transfer problem for the bank, which added morethan 500 branches in six states as part of the merger. Instead of flying tapes from thenew branches to headquarters to transfer customer data, it set up a gateway at theacquired bank's site and fed data electronically into a corresponding gateway connected to2.5TB of DASD. This kept the data online and accessible, while eliminating a manual taskthat could have been very costly.

In moving data from one storage system or data center to another, an IT organizationconfronts three issues: the available time window; the effect of moving data on onlineresources; and the reliability of the data after the move. SANs can also use high-speedconnections and software products to accomplish this such as SRDF from EMC, XRC from IBM,H-XRC and Online Data Mover from Hitachi Data Systems and TDMF from Amdahl.

Other applications that can benefit significantly from SANs include:

Backup and Restore. The sophistication of back-up systems and procedures isdriven by the amount of data involved, the available time window and the restorerequirements of the business (e.g., when the data is needed to continue operations). SANsprovide new levels of performance and flexibility in backup and restore, making itpossible, for example, to backup data from different servers to the same automated tapelibrary. High performance is key to reducing recovery times and SANs make better use ofavailable bandwidth.

Archiving and Retrieval. Archiving data to less expensive, less immediatelyaccessible storage is typically a function of the age of the data and the relative need toaccess it. Archival applications include storing check images and point-in-time recordssuch as customer billing statements. Usually, archive networks are configured eitherwithin a company's multiple locations or in conjunction with a business recovery vendorsuch as IBM, Comdisco, or SunGard. SAN solutions support effective and efficient archivingfrom different kinds of servers to the same storage system. A SAN can provide both local-and wide-area connections and necessary gateway and conversion functions.

Disk Mirroring. Daily backups are often not enough, particularly when theeffects of losing data between backups is substantial or when nonstop data access isrequired. That's when disk mirroring -- the creation of multiple disk images at the sametime -- is desirable. With SANs, organizations can perform database mirroring overunlimited distances, synchronizing the data according to application requirements. Thenetwork provides high availability and fault tolerance, as well as high performance andmanageability.

Shared Data. Beyond shared storage, the ultimate goal of informationmanagement is sharing data -- the extraction, movement, or loading of data betweenenvironments. Think, for instance, of data collected using one platform but needed for anapplication running on another. There are multiple levels of data sharing today and SANssupport them all: network transfer, typically involving TCP/IP; channel transfer, such aswith ESCON; and controller-based shared-storage transfer. This last method can be veryfast (disk to disk) and it allows the use of redundant disk resources. Storage controllersnow permit one copy of data to be shared across SANs by heterogeneous computers.

SANs enable organizations to use existing storage capacity more efficiently, includingcentralizing servers and storage resources that are now widely dispersed. The acquisitionof new storage devices can also be done more efficiently, since the bytes of capacity indifferent storage systems are interchangeable. In addition, SANs enable organizations touse only one set of network management tools. IT managers don't have to undertake costlydevelopment efforts to replicate tools for different operating system environments, whichis the usual situation today.

--Larry Kallhof is Director of Product Marketing at Computer NetworkTechnology (CNT) Corp. inMinneapolis, Minn.

NO BUS TO CATCH

The SAN liberates the storage device; it's no longer on any one particular server bus, but attached directly to the network via a network processor device. In other words, storage is externalized and functionally distributed to the organization.

A SAN architecture makes all storage devices accessible to all servers. A SAN uses a wide range of local- and wide-area technologies, including Fibre Channel, Fibre Channel-Arbitrated Loop (FC-AL), ESCON, IBM's Serial Storage Architecture (SSA), DS-3, ATM and SONET. Whichever technologies are used, the objective is a high-performance subnet with fault tolerance and multiple paths. In effect, a SAN combines the high performance of an I/O channel with the connectivity of a network.

Implement a SAN as a separate subnet. The critical I/O traffic between server and storage should not be blocked or delayed by other kinds of traffic. But building separate physical networks is prohibitive. In fact, many organizations have spent recent years merging separate networks for IP/IPX and SNA traffic and channel extensions. The solution is to provide different classes of service on a shared physical network. Fibre Channel has this capability, as does ATM. And emerging technologies like IP v6 will further enhance this prioritizing capability.

Attach heterogeneous storage devices directly to a network. This requires a special kind of networking device, one that can handle the classes of service that are established and both channel and network protocols. These devices should also provide a fault-tolerant architecture, guaranteed data delivery and data integrity, load leveling for efficient bandwidth utilization, data compression and alternate path routing.

Security needs to be addressed differently. Because storage devices are not protected behind servers as they are in traditional architectures, third-party security, encryption and firewall technologies play a critical role. But each installation has to be examined and security has to be designed to fit the situation. In a SAN world, networking, storage, database, network management and systems management providers will need to cooperate more closely than they have in the past in order to provide complete security solutions.

--L.K.