[mdnsext] mdnsext & SLP

[Garret Peirce <[e-mail]>]

I'm curious why SLP is not a potential solution to the hurdle of 'service
discovery beyond the local link'.

Ignoring client support for the moment, SLP seems like an existing, defined
mechanism for scalable SD.
It seems this draft aims to develop an alternate method to do so.

Finding some history in older IETF zeroconf archives(
http://osdir.com/ml/ietf.zeroconf/2002-07), I'm wondering if the thoughts
to the question from a decade ago would be different today?

>So it seems to me that either DNS-SD has to scale gracefully all
>the way up to huge, or that I will have to bite the bullet and
>build in something tougher like SLP anyway. And then why would I want to
do DNS-SD as well?

As anything else, both this DNS-SD hybrid extension and SLP have their
levels of complexity and choreography.
SLP has hosts advertise their services to a central depository, and the
client must use this depository (SA/DA) to locate known services.   Whereas
in the DNS-SD extension method, the clients use the hybridProxy to actively
search for services. I wonder if the latter might become chatty , the need
for suppression of certain responses and use of LLQ potentially tricky (are
LLQ responses  supported by all common clients?)

Perhaps a current discussion/comparison of DNS-SD/SLP would be a worthwhile
exploration for SD beyond the local link.  I could envision room for both
depending on one's environment, ex. DNS-SD (as it exists) used for manually
controlled entries and SLP for dynamic ones.

Seeing OpenSLP 2.0 was recently released, there is an interesting note
mentioning future mDNS integration - 'Version 2.1 is slated to include
features like mdns integration'.  http://openslp.org/

Appreciate any thoughts as I try to understand the differences of each
approach.

-- 
Garry Peirce
Networkmaine, University of Maine System