SANs: Highest Speeds and Reliability

Updated: April 30, 2009

What do companies want most from their storage technology? Simplified management? Storage area networks (SANs) have that. The highest possible levels of availability? SANs deliver that too. Better capacity utilization? Check.

It's no wonder, then, that an increasing number of large and mid-level organizations, and even a vanguard of small and medium-sized companies, are moving from direct-attached storage to SANs to fulfill their ever-growing storage demands.

A storage area network (SAN) is a set of interconnected drives and servers that acts as a centralized pool of disk storage. These dedicated, centrally managed, secure networks, which generally run over proprietary protocols, use a block-based approach to data storage and retrieval, making them particularly well-suited for environments with heavy block-level data-access requirements, such as databases.

Since SANs are separate from the regular network, disk backup and data transfers can be done directly and automatically without tying up server processing resources. A SAN supports data transfers between computers and disks at the same high speeds as direct-attach units, but because any server can ideally access any storage unit, application performance will improve.

And unlike direct-attached storage, SANs' pooled storage space solves for data managers the long-standing problem of underutilized storage. One final benefit: SAN architecture is redundant, meaning if one server fails, other application servers can still access data.

Up until now, medium and larger enterprises seeking to support their mission-critical, high-volume applications were the companies most likely to adopt storage area networks. And up until now, those companies were choosing Fiber Channel as their network connection of choice. Fiber Channel (FC) is a gigabit-speed network technology employed primarily to transport SCSI traffic from servers to disk arrays. Despite its name, FC can also run on either twisted-pair copper wire or fiber optic cables. In exchange for its superior performance and reliability, companies pay a premium for FC gear--routers, switches, directors and host bus adapters.

However, recent data indicates both those situations might soon change. Studies from Enterprise Strategy Group and IDC show that smaller companies are catching the SAN wave as well. In one such recent survey, for example, some 35 percent of small and midsized businesses said they have moved from direct-attached storage to networked storage, with another 40 percent indicating they were considering such a move.

And such companies--along with their larger cohorts--are looking with serious interest into transporting SCSI data over TCP/IP networks via use of the iSCSI protocol. As a new report from Forrester Research explains, such a setup, sometimes referred to as IP SAN, offers several advantages over Fiber Channel: iSCSI allows firms to continue using the Ethernet and TCP/IP networks they already have in place; Ethernet components are considerably c heaper than their FC counterparts; and, as Ethernet technology improves in the next couple of years to approach 10 gigabyte levels, the performance differential between Fiber Channel and iSCSI storage area networks is expected to disappear, the report predicts.

In fact a Forrester survey of 50 key storage and business-continuity decision-makers at North American enterprises found that 19 currently deployed or planned to install new FC SANs, while 29 had the same plans for iSCSI storage. Even newer figures from the Enterprise Strategy Group: of 511 organizations recently surveyed, 17 percent said they have deployed some form of iSCSI SAN in a production environment, with an additional 20 percent saying they plan to deploy iSCSI within the next 12-24 months. The world of storage area networks, then, belongs to Fiber Channel, but perhaps not for long.