Re: [6lowapp] HTTP and SIP

[Adam Dunkels <[e-mail]>]

Jonathan Hui wrote:
>> Both these arguments were frequently used in the past against IP for 
>> resource-constrained systems, yet today we take IP for granted for 
>> these systems despite the overheads of IP. We need to a pay price for 
>> interoperability and flexibility, but we think it is worth it.
> 
> I'd be careful carrying the interoperability argument towards the upper 
> layers.  IP was designed to be the narrow waist.  Today, that has 
> evolved to UDP and TCP being the narrow waist [1].  We have far more 
> flexibility and opportunity at the application layer than we do with 
> TCP/UDP/IP and we should not be afraid to exploit that.  I do support 
> the argument for evaluating existing protocols/architectures, but more 
> because they are established, vetted, and widely understood, less so 
> because we actually want to utilize the existing technical 
> infrastructure that surrounds it.
> 
> [1] http://tools.ietf.org/html/draft-rosenberg-internet-waist-hourglass

I fully agree with you in that these arguments are stronger than the 
interoperability argument, and the application layer is the place were 
we are free to do what we please, without being constrained from the 
network infrastructure, and that we should use that fact to our 
advantage. (Even though the waist nowadays sometimes is IP/TCP port 80 
or even IP/TCP/HTTP port 80.)

But we shouldn't overlook the interoperability aspects with existing 
application-layer infrastructure such as SNMP, for which there are 
numerous tools and systems available, and HTTP, that has both 
infrastructure (e.g. caching proxies) and deployed systems.

I am not implying that we must necessarily use existing protocols for 
everything in 6lowapp. It is dangerous (and easy) to become too fixated 
with existing mechanisms, particularly in cases where they clearly are 
not applicable. But rejecting them prematurely based on perceived 
qualities and not empirical data is not good.

/adam
-- 
Adam Dunkels <[e-mail]>se>, [phone]
http://twitter.com/adunk | http://www.sics.se/~adam/