James, that's a very good question, from several perspectives. Namely, from the technological and philosophical. Perhaps more influential than the technological will be the philosophical imperative to erase SONET, stemming from religious warfare over the subject which has pitted Cisco against LU and NT in a duel which I believe has reached a near-anachronistic state, but wages on, nonetheless. I say near anachronistic because both sides in this affair have taken measures to meet the other's strong suit, effectively offsetting some of its pointedness.
The impetus to replace SONET in its entirety has undergone a reprieve as evidenced by several factors which I cover below, however, during this past year, but this might only be a respite of sorts, when viewed over the longer term. No one knows which way this will turn out, or by when, at this time. For certain, other line signaling protocols are already beginning to displace SONET for some application types which heretofore were relegated to "on prem" functions, only.
A primary example of this can be found in network attached storage (NAS) schemes over metropolitan distances and beyond, now made possible through the use of FC or GbE over DWDM. Here, the low-loss and high-bandwidth characteristics of fiber effectively neutralizes the effects of distances which heretofore made such applications near impossible, or prohibitively expensive.
Your question was:
"Do I understand that correctly by asking if this Sonet legacy containerization can be done away with once the end user is connected directly by fiber? (Doing away with the need for add-drop muxes?) Or is this containerization not only a function of the physical equipment being used, but also of the non-physical (such as software, etc?)"
SONET ADMs (add-drop muxes) are being replaced by Optical ADMs, or OADMs, which are souped up DWDM devices which do essentially the same thing, only OADMs are capable of multi-protocol handling at Layer One, as opposed to pure SONET and lower order North American/European Hierarchical T1, T3, etc. Again, the other protocols are FC, ESCON, GbE, HIPPI, etc. which makes the newer OADMs a superset of ADMs and digital cross-connects such as those from Tellabs, Alcatel, and others. The latter, interestingly, are now being (retro-)fitted with optical interfaces, as well.
I say that there's been a reprieve to its demise because of a compelling level of recent evidence we've seen in its extended staying power in the "next gen" dwdm and terabit devices, and because of certain tacit, and some outright, endorsements which companies like Cisco have given it by multibillion dollar acquisitions of other companies whose products are based on the stuff.
The most pronounced example of this, of course, was the $6.7 Billion Cerent acquisition by CSCO during the past year. This sits in stark contrast to the previous domination imperative (conquer or be conquered set of mandates) that Cisco had previously postured themselves into with their "old world" schtick.
---
There are two aspects of SONET which we must consider:
- Line handling, and
- Legacy Administration
The first is that quality which I've already discussed in the upstream messages. To expand a bit on those, they consist of timing and synchronization of flows, framing, payload sizing, overhead surveillance bits, etc., all the things that make up the managed container aspect which is responsible for shipping and receiving data at the lowest level.
These parameters needn't go away for any reasons which are deemed imminent or currently apparent, since doing so very arguably would result in the requirement for another similar line protocol to first be established, or chosen from GbE, FC, etc., as a replacement protocol. Such might only prove to be a Pyrrhic victory, at best, unless the new protocol was supportive of some newer upper layer attributes. Such could materialize, say, if a new optical framework were deployed which was based on a different set of assumptions than we're now using.
Again, Cisco and others have already laid down the conceptuals for these new protocols in white papers and long term initiatives, but, to date, these do not present a broad-based real world alternative, and probably wont for several more years, at the very least.
The second set of parameters that I mentioned above -i.e., those which are mainly administrative- tie into legacy operations support systems (OSSes), which are in turn tied to the larger circuit-switching architecture. These are the attributes which are being "nulled out" in the leading edge optical based network elements (very high speed routers, DWDMs, OADMs). This effectively disassociates, or detaches, the newer implementations of SONET framed flows from the framework of the older PSTN architecture for all but the most basic flow handling purposes.
SONET's role may begin to lose share on a percentage basis over time, but it's absolute demise is way off into the distant future. Instead of an all or nothing scenario for the foreseeable future, we'll see increasing diversity and choice being exercised over fiber routes with increasing use of GbE, then 10GbE, some FC, that of the lingering ESCON, and even some newer protocols, perhaps. The introduction of DWDMs, and dark fiber to the enterprise, in other words, will not in and of itself signal the elimination of SONET.
----
I never did get to the restoration part of your original question in reply 824 from the other day. Perhaps later on, it's time to fly.
Regards, Frank Coluccio |
