re: Polatis Plugs On With Optical Switches<quote>So do you believe there is not future for optical telecom? Just point-to-point fibers from LA to NYC and everything else electrical? Does not have much sense just based on electrical reach length of a few meters at 10 Gb/s.
Polatis needs to reduce the size to something resonable and talk about prices.</quote>
Given the fact that most networks in the ground were designed with architectures that rely on periodic OEO terminations, it would seem plausible that they would continue along these lines if the cost of transponders continues to drop. Start-ups like infinera could potentially realize 10x cost reductions in termination costs.
I don't see all optical as a wholesale replacement of transponders in the network,particularly at present costs, as you point out. Unless we can get optics components costs to follow a similar path as electronics (moore's law), it will be cost prohibitive to deploy all-optical where it makes the most sense- at the edge.
re: Polatis Plugs On With Optical SwitchesThe technoloy is similar to Bainbridge Networks. I believe it could be made cheap, since piezzos are not that expensive. Contollers are expensive though, and big. With current size of the box Polatis has no chance. They might think about getting a single multichannel controller and repackaging the box or moving to calient's space, whcih is not very crowdy nowadays.
re: Polatis Plugs On With Optical Switchesbalet said: The technoloy is similar to Bainbridge Networks. I believe it could be made cheap, since piezzos are not that expensive. Contollers are expensive though, and big.
Bainbridge arch is huge- they use cylindrical lenses and larger scale actuators to achieve their beam steering.
Agree- piezos can be cheap, and can be packaged compactly with the right integration, but high-voltage controllers do pose a challenge.
Along with size, all-optical fabrics also have a cost impediment, as typically, the whole fabric must be installed up-front, making them unattractive from a first intalled wave cost perspective. Smaller modules, like waveplane switches can fix this issue, but are a nightmare to integrate and control.
There has been a lot of buzz with hybrid OEO/OOO switch architectures, but these appear to be more academic than anything else, and no one seems willing to step up and fund a development.
This was the basis of my initial comment,in that these architecures, which were driven mainly by cost-savings, will become moot as transponder costs continue to plumment.
re: Polatis Plugs On With Optical SwitchesShaggy, you are right.
Talking to OEM houses about that, Polatis will have hard time competing with other small crossconnect companies, whose products are much smaller and potentially cheaper. If I worked for Polatis I would've tried to play reconfigurable by SW switch architecture.
re: Polatis Plugs On With Optical SwitchesVent said
'Is Polatis planning a ROADM on a chip like others are thinking about' The polatis MASS technology has small piezzo elements moving ferrules with the fibers inside aligning them to each other. The good thing is it's low loss, it will however NEVER be cheap or possible to intergrate it.
Do other copmpanies have piezzo elements controling things such as tuning cavities? Are they not comercialized already...If not then the biggest transponder whore will have to reconsider the primary supplier for the lasers for the UT-2 style of transponders??
re: Polatis Plugs On With Optical SwitchesSo, who is the biggest transponder whore and who is their supplier? Also are you talking about Cisco transponder standard?
re: Polatis Plugs On With Optical SwitchesOptical switches remain expensive and free space piezo tuned switching (i.e. Polatis) will never be cheap to build.
Promising optical switching technologies would have to be integrated optics affairs with low insertion loss, low power requirements, and preferably capable of high speed 10 ns switching.
Such switching speed in conjunction with other architectural/technology advances would allow for transparent (no OEO of the payload) fine grained routing/switching. The two most promising paths are optical packet (very long term technology) and burst switching (near term technology).
All optical wavelength/circuit switching would seem to be losing its advantages as networks are rapidly becoming data centric. I assume that data centric networks benefit more from distributed fine grained routing than from static large bandwidth wavelength architectures.
Now none of this will be adopted until the economic benefits of OOO adoption clearly and continously outweigh the improving economics of OEO based networks and the risk of making a migration to OOO.
The key question then is: When will OOO network economic benefits clearly and continously outweight the economics of OEO networks?
What measurable factors will make it clear that such benefits exist?
If OOO networks were free then the benefit would exist but this is not a practical expectation. More realistically one should look for an economic tipping point such as:
If OEO networks end up being incapable of keeping up with revenue producing traffic demand and if that unserved revenue represented more net profit than the cost of adopting OOO approaches, then a clear and continous stimulus to adopt will exist.
The question is then, will such a tipping point ever occur? Does anyone believe that the networks moving towards such a tipping point in the next 5 years and if yes, please explain why.
Thanks for your insightful input. My vast perceptual acuity also enables me to note that you failed to provide any meaniful answer to my question...