To:
Patrick Greenwell <patrick@stealthgeeks.net>
Cc:
ietf-provreg@cafax.se
From:
Dave Crocker <dcrocker@brandenburg.com>
Date:
Wed, 29 Aug 2001 16:03:55 -0700
In-Reply-To:
<Pine.BSF.4.33.0108290903090.1347-100000@rockstar.stealthgeeks.net>
Sender:
owner-ietf-provreg@cafax.se
Subject:
Re: Message Pushing and TCP Transport
At 09:11 AM 8/29/2001, Patrick Greenwell wrote: >As the number of registries and registrars scale, it might be much more >appropriate to have those registries PUSHing messages only to those >that have messages in the queue rather than having the registrars blindly >polling each registry on a timed interval. Patrik, this is establishing a very dangerous pattern. Besides agreeing with a suggestion from Scott, I'm now forced to agree with your observation. If a registrar is inactive, it should not be keeping a connection in place, wasting server resources. In fact, the idea that the push could include having no connection in place, but have the *registry* establish the connection to the registrar, is not something we have discussed (as far as I have seen.) As you note, however, there is a very strong scaling argument for that sort of mode of operation. (Back in the days of doing disconnected email relaying, I called this "mixed initiative.) thanks! d/ ---------- Dave Crocker <mailto:dcrocker@brandenburg.com> Brandenburg InternetWorking <http://www.brandenburg.com> tel +1.408.246.8253; fax +1.408.273.6464