Hi Gary,
The permutations that we can expect to see are almost guaranteed to be too diverse and numbered to allow us to accurately regard them in general terms. For sure there will be some instances when signaling is achieved through a "control plane" similar to the way SS7 does setups and teardowns of calls for voice today. Only, in the IP realm it will be lambdas, as you suggest.
This has been proposed by the ASON initiative with backers who are a part of the OIF, with endorsements from Cisco and Ciena, OIF's founders. And in other instances edge-to-core will be handled via user-to-network-interface (UNI) protocols that include a form of TCP signaling, such as the SCMR approach that they are attempting to gain a following in, in the ODSI model.
Still, in others, we'll see static WDM link conditions from the edge to the core that are over-provisioned at all times, from an optical lambda perspective. These may well employ a form of rate adaptation capabilities. In such an instance the lambda serves as a constant virtual access mechanism to the core (or, transparently "across" the core to another edge point), varying in throughput as a function of traffic demand.
I would look to Gb and 10 GbE and higher to satisfy this requirement between routers in the future, at Layer 2. To some extent we're seeing this today between boarder routers and distribution routers, already, in large MANs.
Or, lambdas will continue to be established and defined at specific channel sizes much like private lines are provisioned today, through the use of optical cross connects (or even manually, forsooth!), with fixed monthly terms and SLA conditions.
I don't have any definitive notions on a singular dominant outcome to solving these requirements at this time, as the platforms are still evolving -- devolving, if you think in terms of a form of SS7 emulation -- and the eventual standards outcome(s) is (are) still unclear, both to me and to those who are sitting on those standards bodies.
This, despite their high levels of confidence today. ASON appears perched to gain favor through the use of a signaling control plane approach, which I liken to SS7 in many ways. While ODSI seems to have suffered a setback due, in part, to its differing approach which is through the use of a TCP-like signaling technique.
I can't help but think of the similarity here, a bit ironically, between SCMR's attempting to use a TCP technique and being rejected in favor of a "control plane", and that of Cisco's earlier opposition that it met during earlier on, when they, too, were offering a TCP alternative to ATM and SS7. What are we seeing here?, one wonders.
I'm glad we had this opportunity to chime off on this issue, because it's caused me to look closer at some of the underlying issues. As in the case that I'm examining with justone on the LMT Thread, where SS7 and AIN/IN are being examined for their place in the universe in future IP nets, these issues surrounding IP edge-to-core are, too, and IMHO, at the "core" of a turbulent storm.
Perhaps it will always be such, going forward, probably even getting more chaotic with time. In other posts here in SI I've sometimes dropped my guard and suggested: "When the storm subsides, then we'll be in a better position to examine the merits of each case."
But the reality is, I'm quite certain --now that I am thinking more lucidly -- that the storm will only grow in intensity over time, and it probably wont subside at all. Leastwise, not in any form of equilibrium that we are familiar with.
Rather, it will be cause for us to witness different implementations on many interfaceable fronts, which are very likely going to be along vendor or competing consortium (dare I say "paradigm"?) lines. How many internetworking paradigms do we have room for? I guess that's what makes an internet an internet. At least in the original context of the term.
Which purpose was to facilitate: The ability to establish communications between end points (nodes) on dissimilar networking platforms. We tend to lose focus on that point, because to a large extent the Internet has become so homogenized.
For the latter we can look to Cisco as being one of the principal agents of commonality (a.k.a, standardization), since over 75% of the 'Net is either Cisco, or must be compliant with Cisco, in order that it functions at all.
Thanks for getting back to me. It's a great topic.
FAC |
