--On Tuesday, 21 November, 2006 22:07 -0500 Keith Moore
<moore@xxxxxxxxxx> wrote:
> actually the SRV record can easily be a step sideways or
> backward if not carefully used. let's see...it slows down
> session establishment; increases the load on DNS; increases
> the potential for failures due to misconfiguration, link
> congestion, link failure, or server failure; and in some cases
> gives ISPs yet another hook to impose walled gardens on their
> customers, decrease the transparency of the network, and thus
> impair the ability of the network to support new applications.
And, to add one more observation to Keith's list, unless one is
extremely careful, especially when considering using SRV to add
support for protocols that were defined without it, one also
risks recreating all of the problems that caused WKS to be
deprecated. In other words,
* If there is no SRV record present, can I assume the
service is not supported? (No)
* If there is an SRV record present, can I assume the
server is supported and available? (No, not that either).
> there are cases where SRV is useful, but it's a big stretch to
> call it a great leap forward.
Agreed.
john
_______________________________________________
Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf