On 3/16/2004 6:23 PM, Keith Moore wrote:
something that works 90% of the time is unacceptably unreliable.
something that works 99.999% of the time for 90% of users might be
acceptable, provided that the 90% contains essentially everyone with a
reasonable expectation of having mail autoconfigured.
IP does not "work" when there is no data-link layer. ICMP does not "work"
when firewalls filter it out. These same restrictions (and a whole lot
more) can also keep SRV from working in situations that otherwise appear
to be normal conditions. BUT, they also break EVERY OTHER config service
too. Moreover, SRV "works" in more situations than all of the others. I
mean, DHCP is pretty cool but it doesn't "work" when I'm dialing up from
an airplane and getting "local" PPP (that changes every 300 miles) and no
localized proxy gateway to my home server.
Here are some of the problem areas to consider. DHCP leases are bound to
addresses, but not everybody needs an address. LDAP and some other
app-layer services generate too much data to be useful over small/long
pipes. Some users and services expect config mechs to work on a site-local
basis, while other expect mechs to work on an authenticated user basis.
The "ultimate" configuration service MUST specify local file-formats and
locations to ensure that all applications correctly share local options
(like local mail folders), and all filesystems must be accessible from
everywhere. Blyeah.
SRV works in most situations. It's the lowest common denominator. It's the
cockroach not the gazelle.
--
Eric A. Hall http://www.ehsco.com/
Internet Core Protocols http://www.oreilly.com/catalog/coreprot/