SRV resource records and web browsers

[Jonathan de Boyne Pollard]

BM> The only advantage that the suggestion of "fixing CNAMEs" has 
BM> over this is that there are far fewer servers than browsers, 
BM> so it could be easier to phase in that sort of change.

It's not quite as clear cut as that.  At least one popular web browsing 
software has the capability of being automatically upgraded with updates that 
are published by its manufacturer, which a significant number of its users 
take advantage of (and have even been pressured, by events, to take 
advantage of in recent years).  It would not be _that_ hard to roll out to 
a reasonably large number of users an updated version of that browser that 
used "SRV" resource records.  No DNS server software manufacturer has an 
equivalent mechanism for rolling out updated versions of its DNS server 
software.

BM> _www._tcp.www.mycompany.com. IN SRV 0 1 80 server.hostingcompany.com.

_www-http._tcp.example.com. 86400 IN SRV 0 0 80 hosting.example.net.

by strict reading of STD 2 and RFC 2782.