Overview

Today s large switch market in ATM engenders some honest confusion. As both the number of players and the volumes shipped are quite small thus far, there is a Janus-quality to the promotional literature. One face courts the dedicated private network buyer and describes switches for the core of large enterprise networks. The other woos the carrier market and pitches products for the central office (CO). They are, of course, give or take some hardware and software options, the identical product.

One path from this convenient confusion is to describe the large ATM switches functionally in terms of their capabilities. As a group, large ATM switches usually display the capabilities (or limitations) below:

• Offer huge aggregate bandwidth (tens of gigabits) capable of incremental deployment (scalability);

• Support optical carrier (OC)-n and STMn media only;

• Provide redundant (1 + 1) provisioning for fault tolerance on big-bandwidth bundles;

• Claim carrier-grade reliability (redundant everything);

• Process ATM only, no legacy protocols;

• Support both UNI/NNI views;

• Offer software packages to perform user accounting and chargeback;

• Represent big ticket items with relatively few vendors (Nortel, AT&T, Fujitsu, Alcatel, NEC, Siemens);

• Are available with a choice of network management systems, one proprietary, as well as one based on either the CMIP or SNMP standard;

• Compared to edge switches, contain less total software lines of code (LOC) with a slower release rate.

Further, there is an air of unreality when dealing with the applications that these large switches will be handling. It is not evident that any of the public...