Hi Jim,
Allow me to preface my reply with some housekeeping. My use of the term "MSP" was in deference to 4finger's use of the term, but in reality the providers we're talking about here, who deliver Ethernet over the WAN to end users locations, do not fit this category. Metro SPs are companies like MFN, the cable cos, the ILECs, private last mile fiber trenchers, etc. These Ethernet carriers are actually specialized data carriers. I don't have a specific name for them that I'm comfortable with, yet.
Secondly, my allusion to the ILECs at some point adopting Ethernet as a service offering was not an endorsement or recommendation, rather, it was my observation of things to come. At the present time, companies like MFN stand a far greater chance of becoming the carrier of choice for this space. Not so coincidentally, however, we find that VZ is a principal investor in them, as is (I believe) SBC at this point. At the very least, they both have partnering and fiber swap/colo agreements with MFN, in addition to financial ties. So, here we may be seeing a subtle form of responsibility shift taking place, if not a form of implied delegation.
And thirdly, my slant is far from siding with the status quo, where dark fiber is concerned. In my previous two posts I cited what I believe will take place, not what I'd like to see take place.
Refer to the following when you get a chance, to get a feel for where I think we should be going. It describes how some Canadian dark fiber initiatives are beginning to take root, and where they plan to go with them over time:
THE COMING REVOLUTION IN DARK FIBRE NETWORKS"
Andrew K. Bjerring and Bill St. Arnaud
Can Canada be a World Leader in the next Internet Revolution?
canarie.ca
------------
Your question:
" ... it seems to me that the problems of optical switching are such that it can only be performed at a higher level: at the wavelength level, on fiber. Switching occurs by (lambda?), not content: is that correct?"
At the present time, yes, that's correct. Content switching can occur at lower speeds through the use of Layer 5 through Layer 7 switching. But at optical speeds, technology has not yet reached the point where packet headers which can be similarly read, route lookups peformed, and forwarded. Instead, at the optical level, content is lopped onto a lambda, and the entire lambda is switched, or routed.. depending on the vendor's spin.
"I'm assuming that such switching can occur because data is 'pre-sorted' before it is inserted into an appropriate wavelength."
Precisely. Many bundles of like data are grouped together and assigned a label and then superimposed onto a lambda, as in Multi-Protocol Label Switching. Only, in optical routing (which is still really layer two switching at this point) they call it something else.
Like, Wavelength Routing Protocol (WaRP), or something similar. It seems that every vendor has their own favorite cutesie "tag" for this. [Hi Curtis!]
"If I'm understanding what I'm reading correctly, the problems of reading packet header information, and switching packets, cannot be done at the optical level: it is o-e [-o], or, the data stream itself must be routed to a slower network, where the necessary switching, and transfer, continues, perhaps with re-insertion of old data, or insertion of new data, into an optical stream, or to substreams of slower, dedicated data, such as video, or voice."
All of which proves that you were fibbing in your opening line:
"My limited understanding of the subject reduces.." [smile]
Okay, now you are "looking for a problem:
"I'm still trying to figure out how such switching would occur in an Ethernet sub-network, though SONET seems a little clearer."
Keep it simple. Ethernet, in the context that we have been discussing it here and over in FCTF, is a point to point link. Switching and routing do not take place on the line. Ethernet payloads can be "switched" at Layer 2 at the GbE switch, or they can be routed at Layer 3, either in some Gb switches that are so provisioned for Layer 3, or more typically, in traditional routers.
"The extension of optical networks, in the Metro model, ending with FTTH, seems prohibitively difficult: how can content information possibly be switched at optical speeds? FTTH, a concept I once threw around with abandon, now seems laden with huge obstacles."
There is no single definition of FTTH that we could both agree on unless we specify one. From discussions on the nFCTF we found that Optical Solutions would employ a PON (passive optical network, akin to full service area types employing ATM), and World Wide Packets would employ a standard Ethernet Switch to every home using discreet fibers.
Two vastly different models, both from the perspective of complexity and economic gain.
The first would give me cause for concerns, similar to your own. But the second is straightforward and simple, although lacking in some controls that some providers want to have.
The WWP model is represented by a switch at the head end or field node that feeds individual fibers to each home. On the back end of that switch is where content is switched, and it is done in the electrical domain, not in the optical domain. It only becomes optical when the lasers on the line side have been modulated with the electrical one and zero signals coming off the backplane of the switch.
"I realize that I have opened a constellation of subjects here."... hey, that's great.
"... can you suggest any books that would get me
started (at least)? I am particularly interested in an overview of existing network architectures, switching and routing, and how future optical networks might evolve/coexist with legacy networks."
You appear to have a better grip than you give yourself credit for. If you would like to start off with a conceptual view of what wdm is (what the market calls DWDM, that is), I'd recommend going to the iec.org site where some basic tutorials exist. For DWDM, try:
iec.org
Many other topics, including the ones you mentioned, are also covered there, as well. I'll peruse our library on Monday and see what strikes me as a good starter text to meet your need. Off the top of my head, I'd recommend anything that Bill Stallings wrote in the context you described. Lets do a search on Stallings at B&N and Amazon, for starters. Here's an interesting find. One that was published in Jan. 2001, and received a five star review! <g>
Business Data Communications
by William Stallings, Richard Van Slyke
amazon.com
$86.00 Availability: Usually ships within 24 hours
Hardcover - (January 15, 2001) 659 pages [ _grin_ ]
FAC |
