Do we need a Swiss army knife as a communication protocol?
this is a question I raised on my blog.
Using fewer protocols means less hassle to synchronize between them all, but this comes at a cost of protocol stack size and complexity.
Do we want a protocol that does it all - a Swiss army knife, or should we have specialized protocols for each task - penknives?
Good Answers (1)
As far as we know, this was the purpose of SIP at the beginning. Open enough to do everything. But there are difference between Standard Institut and the business world.
What happen to SIP would happen to the Swiss army knife protocol (SAKF). Thinkers will love it and promote it. Consulting companies will create a big buzz around it, Telco will want it, but Industry will adapt it to its needs. If the protocol can do everything, then, it can specialized (I take a little of security, lots of bandwith-management, take away the options, ...).
To my mind, today's buzz is IMS. Everyone try to self IMS as THE-OPEN-PLATFORM-THAT-DO-EVERYTHING-INCLUDING-DIET-COFFEE. Reality is different. Some companies are doing some parts of it and specialized it (meaning not interconnect with other vendors from scratch), other are doing specific part following the standard point by point but then can not interconnect with the first companies.
So, in my mind, better to have specialized protocol, then the game is clearer and the telcos are not mislead by some artificial buzz created by Thinkers and Accenture-like Corp.
More Answers (3)
Trevor B
Software Engineer at Aria Networks
Best Answers in: Wireless (2), Green Business (1), Telecommunications (1)
Hi Tsahi,
I've done a lot of work in the last few years on compressing protocol stacks for use on smaller platforms. This also includes using the same stack for several protocols which leads to a welcome reduction in the resource needed to handle the resultant functions.
I believe this is deinitely the way to go. My work on FemtoCells has also used much technology on stack compression.
I'm really interested in this stuff so drop me a line if you want more info.
Trevor.
Links:
A single, general purpose protocol that does everything (GPPTDE) will require more processing power at the endpoints, more bandwidth, and will be optimized for nothing. As long as you have unlimited bandwidth and infinite resources (e.g., memory, processing power) at the endpoints, it should work adequately most of the time. If you don't, well, then you might want to consider something more suited to your particular purpose. I vote for pen knives.
Moshe H
VOIP and IP Networking Expert
Best Answers in: Telecommunications (5), Computer Networking (3)
Hi Tsahi,
I am a firm believer of Service Oriented Architecture and granularity. Similarly Object Oriented architectures also work in disjoint parts that communicate with each other.
Since I don’t know when there has never been a protocol that defines and solves all. There are frameworks that solve issues. H.323 is a framework with a bunch of protocols, so is SIP and so is IMS. Of course there is a core protocol in each but a lot of other protocols without which the core protocol can do nothing.
H.323 is much more closed than the other two.
Instead of a do-it-all (Swiss Army knife) there is nothing wrong to have well defined frameworks that might include many protocols and standards. But of course there are some tradeoffs such as the bigger application sizes and debugging is more difficult and need a wider expertise.
Let us take video conferencing solutions which are your company’s (mine also in the past) niche. Each of Video Conferencing solutions involves either H.323 or SIP which in turn involve many other protocols. There is no Video Conferencing solution/protocol that packages its own signaling, management and packet transfer protocols.
Do-it-all protocols might have a much smaller footprint than framework protocols but their down side would be to fix any bug or change one small thing you need to open the whole application, correct it and then recompile and distribute it to the systems.
On the other hand a well defined Framework Protocol in which the interface between different parts is well defined a change in one of the parts does not necessarily means a full overhaul of the full application.
The bottom line is a flexible solution which is open, can change its behavior on the fly (late binding) and can add new services easily seems better suited than a do-it-all application in which everything is defined and hard coded before that.
That is having specialized protocols for each task has many advantages and more suitable to the very fast changing world of telecom.
Moshe Haviv
