Storage area networks: hype and hope
Over the last two years, many suppliers have defined the storage area network (SAN) architecture as a group of storage systems and servers that are connected together via a Fibre Channel hub or switch. Gartner Group`s Dataquest believes this definition is too exclusive and undercuts the potential benefits of this evolutionary technology.
By Tom Lahive
Instead, Dataquest defines a SAN as a cluster of storage systems, which may or may not include servers. The cluster is connected via Fibre Channel, SCSI, ESCON, or other interfaces to at least one storage network device and is bundled with software that manages all or a portion of the data flow throughout the environment. No longer is a RAID subsystem or tape library attached to a single server.
The ultimate benefit of a SAN is its flexibility. A SAN enables users to share storage and server resources--not only within the SAN, but also across the LAN and WAN. While channel cabling, hubs, and switches are key technologies that are easing the mainstream adoption of SANs, the software layer will ultimately deliver storage euphoria.
Dataquest believes Fibre Channel hubs will be the dominant SAN technology in the Windows NT space for the next 12 months. At $300 per port 1999 (OEM prices are listed in table 1), hubs are less than 25% the cost of switches. Most NT environments will only require six ports. Looking beyond 2000, the price difference between managed hubs and switches will begin to converge and switches will quickly become the dominant architecture in terms of revenue and units.
SAN component revenue forecasts were determined by supplier interviews. Dataquest considered the following factors to create its SAN forecast:
- Current SAN component shipments were compared to disk array and server unit shipments.
- Clustering requirements were also considered to determine the average number of ports per SAN shipment to overall unit shipments of disk arrays and servers.
- The operating system mix was compared to switch and hub concentration.
Dataquest believes the SAN market size in 1998 is $371 million, growing to over $1.7 billion by 2002 (see table 2).
So, why all the hype over a $1.7 billion market? After all, OEM spending on SAN components will only equal about 4% of RAID spending by 2002. The answer lies with the ancillary technologies and benefits that will ensue. SANs will fuel new technologies and reshape the server environment for the following reasons:
- Storage software will explode because server-independent solutions are network-based and have agents distributed on SAN equiment, RAID arrays, and tape libraries. Software companies will receive more licensing fees with the increase in nodes.
- Network-attached storage (NAS) functionality will become SAN-based. Users will be able to access capacity and create backups without having to go through servers. The need to purchase more servers to act as file-sharing devices may be negatively impacted.
- Storage services and outsourcing will evolve because the installation and configuration of SANs will require specific expertise. Outsourcing companies and RAID suppliers will have to develop new organizations that help users configure storage applications.
- SANs will act as bridges to LANs and WANs. Leading networking vendors will deliver solutions that integrate LAN/WAN switches with SAN devices, adding credibility to the SAN vision.
- The server will become a peripheral to the storage environment as I/O functionality is distributed to the SAN.
Tom Lahive is a senior industry analyst, specializing in server storage and RAID, at Dataquest, a market research firm in San Jose. Lahive works in the company`s Westborough, MA, offices. He is a regular contributor to InfoStor. For more information, visit www.dataquest.com.