re: Juniper, Telefonica Demo IP Optical Coordination
The need for the connection based (optical, or rather, digital) transport networks to adapt their capacity allocation to the (IP) packet traffic patterns has existed in major service provider networks at least since the IP traffic started dominating the traffic loads.
GMPLS has been proposed as the solution for nearly just as long - from the late 90s on.
But why has GMPLS in 10+ years delivered an economically working solution?
For one practical reason, GMPLS controls the connections via client/server type software processes, and such control loop delays are orders of magnitude longer than packet traffic bursts (possibly, single packet transmissions) based on which the connection capacity control should operate to be useful.
Current GMPLS for 'adaptive networks' are kind of like air conditioning system that checks the temperature twice a year, in January and July, delivers the adjustment with a half a year delay, resulting heating in summer, and cooling in the winter.
Will this Telefonica-Juniper project make the difference this time, and how?
re: Juniper, Telefonica Demo IP Optical Coordination
The reason why GMPLS will make a difference now is that only recently the optical toolset is available to allow the optical flexibility that GMPLS can control. I mean Directionless and Colourless ROADMs as well as robust transmission mechanisms at 40G/100G. I agree that without these tools, GMPLS is of very limited use in an optical network.
The need for the connection based (optical, or rather, digital) transport networks to adapt their capacity allocation to the (IP) packet traffic patterns has existed in major service provider networks at least since the IP traffic started dominating the traffic loads.
GMPLS has been proposed as the solution for nearly just as long - from the late 90s on.
But why has GMPLS in 10+ years delivered an economically working solution?
For one practical reason, GMPLS controls the connections via client/server type software processes, and such control loop delays are orders of magnitude longer than packet traffic bursts (possibly, single packet transmissions) based on which the connection capacity control should operate to be useful.
Current GMPLS for 'adaptive networks' are kind of like air conditioning system that checks the temperature twice a year, in January and July, delivers the adjustment with a half a year delay, resulting heating in summer, and cooling in the winter.
Will this Telefonica-Juniper project make the difference this time, and how?