Intel lays out network stall
Cash-strapped telcos need clever kit
Intel Developer Forum Equipment manufacturers should not expect a recovery in the telecommunications market even in the medium term. Instead they should concentrate on developing kit which helps hard-pressed telcos save money.
That's the view of Sean Maloney, general manager of Intel's communications group, who said today that profits for telcos are dropping to 1997 levels, even though they are dealing with an explosion in traffic volume.
Although capacity in the core of the networks continues to rise this is not matched by bandwidth available to the home. We are highly unlikely to get anything faster than consumer DSL for at least five years, he predicted.
"There is no corollary to Moore's law in terms of bandwidth," said Maloney.
Intel's wants to get product developers using its network processors, and during his keynote Maloney gave an early indication of the features which can be expected in second generation chips based on IXA packet processing architecture.
Second gen IXA chips will be based on xScale and will be "multi-threading microengines" in order to boost the number of networking operations per second.
This will enable processors to perform clever operations like encryption and routing.
During his keynote, Maloney picked out several areas of growth within the networking gloom.
Wireless networking continues to be a bright spot in the market and Intel predicts 802.11b kit will continue to sell well for 12-18 months until products based on 802.11a, which promise improved security features, make a splash in 2003.
Repeating what has now become industry wisdom, Intel forecasts strong demand for Gigabit Ethernet in the enterprise and the use of 10Gbps Ethernet in the metro.
More controversially, Intel predicts that Ethernet for storage interconnects will surplant fibre channel over a three-to-four year transition period (though users will continue to use fibre channel for well beyond this time). Ethernet was never optimised for block data transfers and we're sure Fibre Channel vendors such as Brocadewill be quick to point out the limitations of iSCSI.
On previous occasions FC vendors have told us that using Ethernet for block data transfer is like putting food into a liquidiser and sending it down a narrow pipe before trying to reassemble it at the other end.
However, Ethernet was won out against similar arguments advanced by the ATM crowd, and scores in ease of use and ubiquity. ®
Related IDF Stories
Sponsored: Beyond the Data Frontier