----- Original Message ----- From: <[email protected]> To: <[email protected]> Cc: <[email protected]> Sent: Friday, August 07, 2009 11:27 AM
Well, syslog-over-UDP was acceptable to IESG, and was published as RFC 5426 couple of months ago. Syslog-over-UDP is not the mandatory-to-implement or recommended transport for in RFC 5424, due to both congestion control and security reasons. Syslog-over-DTLS-over-UDP would have the same challenges in congestion control, so probably it wouldn't be the mandatory-to-implement or recommended transport either. But that doesn't prevent it from being published as RFC. <tp> Pasi Thanks for that. One trigger for this discussion was Chris's question, as to whether we could write a syslog-over-DTLS I-D, or whether it would be syslog-over-DTLS-over-UDP or whether it would be syslog-over-DTLS-over-UDPandSCTP .... Obviously an I-D covering more than one substrate could RECOMMEND the one with flow control but, for myself, I would prefer just to cover a single substrate, namely UDP (seeing that as the one most likely to be implemented). So I see a difference from the syslog-over-UDP case in that there is a close alternative to syslog-over-DTLS-over-UDP which does provide flow control ie syslog-over-DTLS-over-SCTP (or DCCP), an alternative that is much closer than in the choice of UDP versus TLS, so we might be encouraged to include a substrate that did offer flow control. Worst case scenario would be to produce a UDP only I-D and then get told late in the day that we have to retrofit SCTP and/or DCCP in order to address the concerns of the then Transport Area reviewers. I appreciate that it is impossible to predict the views of such reviewers in a year or two's time but the reassurance I was looking for was that you thought that such an action was unlikely. Looking forward, I think that Chris's preference would be to have syslog over DTLS with no reference to substrate. Mmmmmmm, I think that that would leave too many issues uncovered, but I think that the question of substrate(s) is the discussion we now need to have, with Joe's views, as potential editor, being of particular interest to me. Tom Petch </tp> Best regards, Pasi _______________________________________________ Syslog mailing list [email protected] https://www.ietf.org/mailman/listinfo/syslog
