OC3 (155 Mbps) for WAN, or fast-Ethernet (100 Mbps) for LAN (FDDI for earlier time), used to be "high-speed" two or three years ago. But today, everything is in gigabit rate in all network categories: WAN, LAN, and SAN (storage area network).
In LAN, even for enterprise backbone, everyone knows g-Ethernet beats ATM, especially on desktop, because of its simplicity, low cost, and compatibility with Ethernet and fast Ethernet. In SAN, it is Fibre Channel that also runs at gigabit rate to be the dominating technology.
In WAN, ATM is still the best choice for its QoS. ATM is also easy to work with frame relay and xDSL. IP over ATM is widely in use and ATM is increasingly playing an important role for the convergence of data and voice networks. But IP over SONET, and even IP directly over DWDM are emerging. Particularly, DWDM is growing so rapidly that 80-channel DWDM carried by a single optical fiber is becoming available from CIEN and LU, and each channel offers a bandwidth of 2.5 gigabit per second.
It is a well-known fact that silicon is limited to sub-gigabit band. It is GaAs (maybe Germanium in future) that is the king for today and tomorrow's broadband networking chipsets.
VTSS is the dominating GaAs networking chipset developer, just like PMC-Sierra dominated the ATM-PHY chipsets. (But VTSS is bigger in terms of market capitalization and revenue than PMCS with higher growth rate and lower price and PE.)
I think ATM still has its days(and PMCS as well). That's why I like NN that is just having a turnaround story yesterday. I have followed PMCS for about four years and watched its phenomenal growth closely. If the management of PMC-Sierra reads my posts (as I believe so), they know how to use my free advice. They seem to know what they are doing.
|
