The Lan market switches up a gear

Lan switches, particularly 100Mbps switches, are firing up the Lantop environment. Ken Mann finds out if this spells the end of the 10Mbps switch and the 100Mbps hub. inter-networking market. According to Dell'Oro research, worldwide sales for this sector grew by just over 15 per cent in 1997 - from $14.3bn (#8.69bn) to $16.5bn - and much of this expansion can be laid at the door of the switches.

However, the dynamics of the market have been unkind to some vendors.

In addition, sales of shared hubs declined during 1997, despite a 111 per cent increase in 100Mbps Ethernet shared hubs.

While customers are happy with this state of affairs, vendors' revenues were driven down as prices of 100Mbps switches plummeted. Prices dropped 21 per cent in the last quarter of 1997, and 100Mbps Ethernet switch sales started to eat into 10Mbps switch and 100Mbps hub sales. These aggressive price reductions are expected to continue, with price erosion expected to outstrip sales growth. By the end of the year, 10Mbps switches and 100Mbps shared hubs may become endangered species.

Last year, the debate at the desktop was when switched Lans would take over from shared hubs and how much bandwidth would be enough - 10 or 100.

With these price drops, switched 100Mbps is no longer at a price premium, and the debate is over.

Initial Lan switch sales did not address the edge of the corporate network, but instead were used to aggregate traffic from other switches and hubs.

These aggregation switches are characterised by high performance, a rich set of features and high-priced functions - such as management software, extra memory and fibre-optic uplinks. High price per port was acceptable, since this could be spread out over a large user population.

As Lan switches are deployed at the edge of the network to connect individual desktops, fewer users are served by each device. Per-port price then becomes a significant factor in the buying decision. The market for Ethernet switches configured for desktop connectivity has shifted from the early adopters to the early majority.

This is why, as Ethernet switch sales displaced sales of shared hubs in the second half of 1996, vendors began to battle for customer accounts.

As Peter Bronson, Asante UK's territory manager, pointed out: "Previously in switch technology there has only been higher density solutions at much higher prices. Barriers to the adoption of Fast Ethernet switching at the desktop have included the high cost, the complexity of configuration and the high port count."

Bronson continued: "At the low end, shared hubs will start to disappear - but there will be customers who still want them. The SoHo market will soak up small low-density hubs for some time to come."

Over the next two years, the workgroup switch market will undergo a similar segmentation to the hub market. Increasingly rich functionality will flow downwards from the high end towards the desktop in order to retain price points and differentiation.

In this gap between the early adopters and the early majority, brand loyalties could shift. This is partly the reason behind IBM Networking Systems' launch of desktop hubs and switches.

Chris Blenkhorn, networking technical consultant at IBM, said: "Previously, our major goal has been to provide an across-the-board solution for major corporates. Below that is the small and medium enterprise (SME) market space. We also believe the IBM brand and channels to market will be a significant factor."

Neil Venus, general manager for UK and northern Europe at ODS Networks, stated: "I'm surprised to see such a muddled vision from IBM. At the low end, IBM is offering a plethora of different boxes with disparate management systems. The product set also has a lot of overlapping functionality. Take management, for example: some is web-based, some clearly isn't."

All things to all people

IBM admits as much. "It is difficult to be definitive. There are several different products addressing the same requirements," Blenkhorn admitted.

Venus concluded: "IBM is becoming a systems integrator rather than a manufacturer - integrating Xylan, 3Com/Chipcom and 3Com SuperStack into its range. IBM should be developing its own network kit rather than relying on others. I haven't the foggiest about what IBM's USP is supposed to be - more of an uninspiring selling point than a unique one, perhaps."

IBM has also revealed a change of commitment in backbone technology. Previously, IBM was committed to ATM in the backbone with Ethernet and other technologies as feeders. This is still in place, but given equal importance is Gigabit Ethernet and Layer 3 switching in the backbone. The new architecture will be rolled out during 1999 - and will be IBM-built.