With at least 30 companies now developing network processors, the market is getting crowded. To distinguish themselves from the horde of lemmings pursuing the 10-Gbit/second market, some vendors are developing special-purpose NPUs.

For example, Trebia Networks is developing a product it calls the storage network processor, aimed at the SAN market. As current Fibre Channel storage networks begin to include Internet Protocol, NPUs become applicable. Trebia has not yet disclosed how it will address the specific needs of the storage market. But by creating a product optimized for storage protocols and algorithms, the startup could quickly gain an advantage over traditional NPU vendors.

While most NPU vendors attack the lucrative edge market, Brecis Communications has developed network processors for higher-volume access equipment. Though Brecis can't match the high prices charged by other NPU vendors, it may well sell more parts.

Other NPU vendors are differentiating by supporting protocols other than basic IP and packet-over-Sonet. Agere has focused on asynchronous transfer mode, while Motorola includes HDLC in its portfolio. Startups Onex and Zettacom support time-division multiplexed as well as packet traffic.

Clearwater Networks is carving out a unique position as a provider of a network-services processor. Its CNP-810 works in conjunction with a standard NPU to provide high-touch services such as network address translation and firewalls. Other startups are taking an even more specialized approach, focusing on Layer 7 and TCP functions needed in Web switches.

NPU vendors that zero in on segments can better meet the needs of that application and set themselves apart from more general-purpose competitors. This differentiation helps attract venture funding.

The downside is that unexploited markets are usually left alone for a reason. Generally, they are niches; even if a vendor can dominate a niche, the total revenue generated can be rather small. If a small market suddenly grows, chances are good that it will attract the competition.

Vendors such as Intel and Motorola are too big for niches. They rely on large volumes to fill their fabs. This leaves opportunities for smaller, more nimble vendors to exploit. It is a classic big fish-small pond question. In the NPU market, the big pond is getting crowded. It's time to seek other ponds, even if they are smaller.

Linley Gwennap is founder and principal analyst of the Linley Group (www.linleygroup.com/npu).