As We are facing enormous challenges in network optimization, why not develop Optimizable protocols for future networking technologies? Please share your views.
Please google "Software Defined Networking"...maybe it will answer a few of your questions! ;-)
SDN gives an abstraction of network to deal network more efficiently. We can consider SDN as a Network Optimization solution. What I feel these kind of solutions can provide medium term solutions for long term we have to reconsider the basic layering architecture of the protocol stack. For example we need a paradigm shift from Source-driven to Receiver-driven approach and host-centric to content-centric services. To handle this paradigm shift we can adopt two strategies 1-Optimiz current protocols to handle upcoming challenges 2-Redefine whole architecture but new protocols must be designed by keeping optimization in mind.
@: M. Bilal: Your above views are good. I may add that , its not feasible to redesign a whole technology/regime of protocols that too for optimization. Optimization of networks in general and network protocols in particular are subject to few application specific parameters. Further more, as an analogy , GPRS was adapted over GSM, as replacing GSM wasn't feasible.
I would like to hear more form you side: its a good topic for discussion
Please be more clear/specific about your optimization term.
Regards:
Majid
The main issue with the concept of optimizable protocols is the backward compatibility.
Just an example: TCP can be greatly optimized, but many strongly optimized (and better performing) variants will never be accepted by IETF because they would result in extremely unfairness when sharing a channel with "older" TCP flavors.
So, the optimization is possible, but can't be deployed.
This is a common pattern in telecommunications. Once a system is in place and have gained wide acceptance, substituting it with a better one is a problem.
Even if you define a protocol stack that CAN be optimized without complete redesign, you'll always face the same issue: fairness will block you, unless you can upgrade all the implementation is a short time, which usually isn't possible.
So, summarizing: good idea, but implementation close to impossible.
In the meantime, I agree that SDN and protocol redesign will be the thing we'll have to stick with. As is, ad-interim solutions and new protocols to "kill" the old, crappy ones.
- Badges
- Science topic