[5237] in www-talk@info.cern.ch
Re: Virtual Circuit protocols => universal access
daemon@ATHENA.MIT.EDU (Simon E Spero)
Mon Aug 22 14:02:54 1994
Date: Mon, 22 Aug 1994 20:01:41 +0200
Errors-To: listmaster@www0.cern.ch
Errors-To: listmaster@www0.cern.ch
Reply-To: ses@tipper.oit.unc.edu
From: Simon E Spero <ses@tipper.oit.unc.edu>
To: Multiple recipients of list <www-talk@www0.cern.ch>
There are very few places where it makes sense to replace a
connection-oriented reliable transport protocol with an unreliable
connectionless protocol. These cases tend to be those where:
1) The typical transaction consists of a single outgoing datagram for
the request, followed by a single datagram containing the response.
2) The typical transaction profile consists of many queries issued to
separate hosts.
3) Operations are idempotent.
4) The loss of a certain amount of data is acceptable.
If a service doesn't fit with this set of criteria then there is very little
point in moving to a connectionless protocol.
Example: DNS
DNS sends a single packet for a query, and recieves a single packet
in return. Typical DNS queries span multiple hosts. Thus the cost of
connection establishment is too high for this type of service.
HTTP generates two datagrams for each query (or at least X-mosaic does),
and recieves a large number of datagrams back in response. The typical
transaction profile consists of many queries to a single host. HTTP
POST operations are not idempotent. The loss of data is not considered
acceptable. Therefore HTTP is not well suited to a connectionless
paradigm.
Simon
