Hi Steve, Thread,
The era of micro-optic-based networking seems to be in the offing at this stage, due to some real world mandates which LUMM appears to be addressing. The issue of assessing DWDM gear for metro, and even LAN and desktop, applications at this time appears, suddenly, to be a real possibility. In examining this space, some logistic aspects, as well the more compelling technological ones, come to mind.
Today's typical DWDM installation is characterized at the physical level as being manually intensive, bulky and difficult to administer in many instances, from a network administration perspective. The greater the distances of the fiber spans, the larger the installations (due to the need to incorporate optical amplification in the long hauls).
When using today's units, changes in lambda mapping require the skills of a qualified technician who must patch, by hand, from one port to another... very often in a maze of orange fiber optic cabling between different cabinets, or racks. Sometimes, between floors.
Reducing DWDM's and other optical networking elements form factors to a place on printed circuit boards, and controlling mapping under software direction, is indeed a liberalizing prospect to those who must run and maintain networks.
Because of this bulk and often skills-demanding nature of existing DWDMs - some can take up multiple racks fully loaded inclusive of manual patch fields and cable management assemblies - they are relegated to only those situations which warrant large investments in order to attain fiber strand gains, or virtual strands. The decisions to select and install these units takes place only after some rather lengthy and sometimes torturous breakeven and ROI analyses, and then, only after the internal politics of carrier- (and increasingly, enterprise-) management issues have been cleared off the table.
Fiber gain is not the only, or even the ultimate utility of DWDM, however. Equal, if not greater, benefits to be derived from lambda manipulation and flow syntheses include switching and routing of lambdas and their derivatives, which will inevitably be found at the local riser and work space level. Here it will be possible to address entire frameworks which are not fully dependent on traditional bulkware (routers, hubs, switching machines and entire Cat-5 cabling distance-defined "equipment rooms") soon, with the potential to obviate many of the prevailing networking techniques.
"Soon," here, may be sooner than many think, translating into a ramp up commencement sometime in the late 2001 to early 2002 time frame. In some special early adopter situations, even sooner, although more crudely at first.
Much of this is being engendered, btw, by the increasing availability of dark fiber by some local cable construction companies (MFNX, FTGX, some local cable TV outfits, and the like) and the almost certain outcome that other carriers will need to respond in kind, over time, lest they lose out.
I can't speak for LUMM, nor do I know if their directions are fully aligned with what I've just described. If they aren't looking in this direction, they should be. From what little I've been able to find on them, however, I suspect that they are headed in these directions. If not them, tho, certainly others are. But I think they are.
And just what are these "directions?" It should be to provide the kinds of functionality that allows the mapping of routes and flows under software controls in ways that are no more difficult than how traditional IP routers are administered today.
Hopefully less cumbersome, at that, through the use of appropriate configuration management software, as opposed to the manual patching techniques that are required today. Just some thoughts, while we await more news from LUMM and several other startups, and the other more established players, who are also addressing these issues at this time. Perhaps the latter are now acting with a little more haste, now that the prospect of a LUMM or a Bookham, and their kind, are now coming into clearer focus on the horizon.
Regards, Frank Coluccio |
