Storage networking for the masses
- 22 May, 2002 10:31
There is general agreement that Fibre Channel solutions are unnecessarily expensive and proprietary. You can blame the profit motive for this, but Fibre Channel vendors may not be completely at fault for cost and interoperability issues.
Providing a transport mechanism for SCSI over a network of storage devices, which is what Fibre Channel is all about, calls for numerous standards to cover the various minute aspects of data networking, including routing, switching, and QoS (quality of service). Such a large set of standards is necessarily coarse-grained, leaving vendors room to fill the gaps with proprietary interpretations designed to boost performance or provide features competitors don't, with the result that solutions from two vendors, say two Fibre switches, can be plug-compatible but incapable of communicating because of incompatible software inside the box.
To make things worse, Fibre Channel vendors have shown little interest in sharing their technology with potential rivals. On the contrary, they have frequently sued perceived violators of their patents. Unfortunately, this trend -- as recent legal battles between EMC Corp. and Hitachi Data Systems Corp. and between Brocade Communications Systems Inc. and McData Corp. prove -- still continues.
Will vendors selling IP storage solutions repeat the mistakes made with Fibre Channel? There is room for hope because IP storage is growing out of more mature technologies and market conditions than Fibre Channel did. The task that early developers of Fibre Channel technologies undertook was gigantic because it involved creating a new networking architecture from scratch and persuading customers to use it.
By contrast, the network infrastructure for IP storage is already in place, and IP players are used to a scenario in which multivendor configurations are the norm. Although IP storage vendors may be tempted to deliver value-add features, the existing IP infrastructure will pressure them to do so in ways that are interoperable with others. In addition, IP storage brings to the table a new group of companies, such as Cisco, that can contribute their IP networking expertise and the knowledge that, when it comes to networking, a proprietary approach does not pay in the long term.
Another reason for hope is that IP storage vendors must compete in a market in which Fibre Channel is king. Their solutions must prove to be a worthy alternative to the dominant technology. Moreover, IP storage vendors will be addressing a potentially bigger market than do Fibre Channel companies, which should allow them to spread R&D costs across a larger base of sales, helping to drive down prices.
A mass migration from Fibre Channel to IP storage is unlikely, but storage managers will be tempted to add less expensive, easier-to-implement IP solutions to their Fibre Channel core as long as it doesn't mean sacrificing performance, scalability, reliability, and management features. For this to happen, however, IP storage vendors must not only deliver on the promise of lower cost and easy implementation; they must also provide easy integration with the Fibre Channel fabric.