Hi Mike,
>>Didn't Hiram hit the nail on the head? Gigabit Ethernet(GE) is a LAN/WAN technology.<<
GE is primarily and foremost a LAN/CAN <campus area network> technology. Its use in Metropolitan Area Networks, or MANs, has been broached, but only in a proprietary way thus far, and I suppose that qualifies it as a kind of WAN technology when chosen, as well. But not primarily.
>>ATM won the enterprise/telco standards business, right?<<
ATM has achieved deep penetrations in the WAN in carriers' internal Core Networks, for the purposes of transporting a mix of other service provider and customer protocols such as TCP/IP, Frame Relay, SNA, etc. But the use of native ATM over long distances (as opposed to access pipe aggregation in the last mile) by large enterprises in the WAN is still lagging considerably, to the point that carriers sometimes find themselves short on being able to address customers' long haul ATM needs when they arise.
In other words, demand has not been sufficiently demonstrated for carriers yet to provide a robust service mix centering on WAN ATM at the customer application level. Yes, they can often provide ATM for customers who demand it, oftentimes on a made-to-order basis, thus far, but not with the normal levels of redundancy and network management capabilities that they can for, say, Frame Relay or ordinary T1/T3 services. ATM has just not been productized and disseminated in the standard offerings to that level, yet. And who knows if it ever will (where end user native ATM traffic is concerned)?
Many enterprise initiatives which were once intended to use ATM in the past are now re-focusing on the use of virtual private networks, or VPNs, using TCP/IP. In turn, the TCP/IP traffic may be a tenant on a larger ATM core network, as in IP over ATM, but not from the user's perspective. From the user's perspective, it's still only plain old TCP/IP, despite the carrier's choice to transport it over their ATM core network.
>>I mean LAN/WAN buildout isn't dead, but near it. Even if there are LAN/WAN upgrades, it is to 10/100 ethernet speeds.<<
That's true at the workgroup level, but it will not remain the case at the server level for very long.
>>Very few are even opting to go with GE speeds in the LAN/WAN. It's going to be viable, but not for at least 3 or 4 years will we see widespread use. It's bleeding edge technology now.<<
It may still be bleeding edge on copper UTP to the desktop, and as Bernard indicated, desktop CPU processor speeds are the primary cause of this. But this no longer is the case in LAN/CAN backbone situations where GE is used to link concentrators and server farms, where fiber deployment is the medium of choice. And with GE over lambda now becoming viable, I can see where dedicated networks for individual enterprises are now feasible at the GE level, especially in densely covered metropolitan areas where dark fiber providers can facilitate this quite readily today.
I have a client who is sending IBM SNA traffic from Manhattan to Brooklyn now over an MFNX link that also supports Fast Ethernet using a 2-WDM scheme. They will very likely upgrade this arrangement to 4-WDM and their Ethernet to GE in the short term, since it is now doable without much bother.
>That is why MRVC stock has been floundering for years, isn't it?<<
I cannot attest to the stock floundering for this reason.
>>Bernard, you mention that GE runs over copper only if there is superior installation on the twisted pair. So this eliminates GE as a last mile solution because the copper in the ground, buried by the telcos over the last 100 years doesn't qualify does it?<<
Correct, the underground stuff is both too long, and does not meet the required longitudinal and termination characteristics of Category 5 cabling.
>>... but I just didn't think any investment in GE was wise. Maybe in a year or two, but not now.<<
That depends on the circumstances you find yourself in, and the topology has a lot to do with it also. If you are designing a campus backbone or a building riser backbone, GE fits quite nicely. If you are planning to upgrade your accounting department's LAN, then no, you are correct in your assessment.
Re: MRVC you say:
>>I felt it was because MRVC has nothing much to do where the real money is being spent, the enterprise/telco upgrades.<<
Maybe it is simply that they are forward looking, and they are planning to make a splash just over the horizon, as opposed to entering a market that is saturated today with commodity players. I don't follow or necessarily endorse MRVC, per se, but their direction, like LVLT's, is right on the money, as far as I am concerned. Like others in their class of offerings, they are focusing on the upcoming bandwidth-rich paradigm that may just leave today's other commodity players sucking up dust when all is said and done.
Regards, Frank C. |
