Stash,
>>You seem to have a handle on the technicals of Photronics<<
I have a good handle on public carrier/SP optical networking from the perspective of being a building blocks type of architect, but I leave the deep science to some associates both here and in work.
See the SR thread here in SI for an idea of what I mean by that. I take a back seat to the scientists there, and enjoy every minute of the research they send me off to after every post. Thanks for the acknowledgement, just the same.
>Can you clarify a couple of things for me? My basic assumptions... Cien and SR are throughput enhancing along a single glass fiber ie. Straight line technology?<<
In the case of CIEN, it appears all very linear to me, if that's what you mean. With SR's claims, I don't fully understand their approach 100% yet. But some of that is intentional on their part, I believe, and not just my lack of an adequate foundation in theoretical physics.
>>TLAB touts their digital cross connect switches...Are they fully optical or do they need to convert from op to electrical signal?<<
I don't mean to be evading your question, but it's all a matter of penetration into the box, how granular they can make their streams, how integral into the box they can continue the optical processes, and where you want to establish a point of reference.
Take the 5500, for example. They use a standard SONET hierarchical structure for both their electrical synchronous transport signal [STS-x] and optical carrier [OC-x] interfaces.
A 155 Mb/s stream can be presented to the device (i.e., to the the TITAN 5500) in both the optical fiber format on the OC-3 interface, or at the electrical [coaxial] interface at the STS-3.
When it arrives as OC-3 on fiber, it must at some point be converted either externally or internally to the STS electrical levels. In that regard, the answer to your question would be, YES. It needs to be converted.
OTOH, in the past, this conversion process was exclusively handled externally, whereas today it can be done internal to the TITAN, obviating the need for an external conversion device (mux-demux assembly). So in that respect, the answer is NO.
Again, marketing literature notwithstanding, the answer depends on the agreed upon point of reference and the nomenclature you choose to abide by.
>>Does this slow through put...<<
No, not at the level of the original signal, in any event. But the overall achievable speeds in this framework are limited due to the need for signal handling in the electrical domain. But that is not what you were asking, or at least I don't think it was.
If you are talking about at a given OC-12, say, would the signal be slowed down, then the answer is no. If you are talking about total achievable top end bandwidths, then the answer is yes.
>>.. can they be remotely configured in the event of a physical problem w/ the fiber...cut fiber to prevent a problem as AT&T had recently?<<
[[As an aside, I am aware of a major fiber cut that MCI had recently in Utah, but not T. Which one was that?]]
Getting back to your point, yes to varying degrees, all present day opto-electronic systems are remotely configurable. And the more integration that exists in the optical domain, the easier it would seem that the configurations can be achieved. A noted variable that exists and what differs from system type to system type, is the time lapse to effect changes. System restoration times, in other words, vary greatly between purely SONETized systems, and those which incorporate DCSs.
When a pure SONET ring is cut, it restores itself within a few tens of milliseconds. However, when the traffic that rides over a SONET Ring is picked off and groomed by way of a series of digital cross connects, then that's a different story. DCSs may take anywhere from a few seconds to a few hours or more to rebuild routes due to inherent speed limitations in both their own architectures and the dependent architectures of the networks they are a part of and serve.
Often, when services are provisioned through DCSs, they sacrifice the native SONET self-healing capabilities that would otherwise exist.
>>What happens at the router ie. w/ CSCO....Optical to elect? <<
For traditional 7200s and 7500s, say, optoelectronic conversions take place at the port level when fiber is used for the interconnect. Coax is often used, however, when the connections are local, and those are electrical from start to finish. For the fiber interface, it's simply a media and format conversion,. But once those signals hit the router's internals, as far as the router is concerned, these are purely electrical signals at this time.
>>Does anyone have in the field or in Lab testing a fully optical switch or router?<<
That's a tough question for me to answer, strictly because there are no established conventions or nomenclatures in place to use for this purpose. There are some companies who claim that their optical routing can take place at the multi-gigabit levels, and actually avoid traditional hierarchical formats altogether. Some of these companies who have the most iconoclastic approaches, are in the R ad D stages, and are only one stop off of the chalk board. Upon inspection, though, they all seem to be doing Layer 1 and sub-Layer 1 adding and dropping of wavelengths.
This is, of course, a form of routing, but it is not consistent with what we generally associate with Layer 3 routing in the 'pure' IP sense. Here's where the new paradigm of optical networking collides head on with the established OSI Reference Model that has been used for the past 18 or so years to satisfactorily define network element responsibilities.
If we take that sub-Layer 1 beam, for example, we can affix a coded label to it, just like we assign IP addresses to packets. What does that make the beam? Does it cause the beam to become a sub-Layer 1-enabled Layer 3 optical packet? These conventions have not been spelled out with any consistency yet, to the best of my knowledge.
>>This was quick, off the top of my head..and maybe not well thought out, however, if you can make anything of these basic questions it would be appreciated<<
Hey, these are all very good questions! You should not feel in any way overly unenlightened by all of this. Some of the best minds where science meets marketing meets engineering are struggling to establish practical tools and interfaces for these emerging technologies, so that they can be understood and used in some kind of consistent manner that follows a natural migration path from those of their entrenched predecessors.
I welcome comments and corrections. If someone would like to elaborate on what actually takes place at the optical interface and beyond, in the Titan, I would appreciate that, too.
Regards, Frank Coluccio |
