More realistically: Call transfer scenarios can happen where "transfer" functionality is provided by a routing gatekeeper, meaning that the transferred endpoint A (the one involved in both a pre-transfer call to B and a post-transfer call to C) does not notice any change in terms of a new call coming up (this is permitted by H.450.2 in circumstances where A does not speak H.450). It will however see (and take part in) fresh H.245 negotiations. Now suppose A and B both support tunnelling and C doesn't, so the initial call from A to be used tunnelling. Post-transfer, either the call needs to switch over to non-tunnelled H.245, or the gatekeeper gains a whole load of extra work it really didn't want.
Chris
Pekka Pessi wrote:
Paul Long Plong@SMITHMICRO.COM writes:
Why would an EP ever want to switch from H.245 tunneling to a separate H.245 connection? What possible use could there be in such a thing?
The H.245 might be handled by a different box than the Q.931? I think Paul Sijben from Lucent presented a scenario, where Q.931 was handled by a MGC and H.245 by MG. (He also presented quite many reasons why nobody would really implement such a scenario.) Another possibility would be a distributed MC, where the Q.931 is handled by one box and H.245 by other ones. Pekka Pessi
-- Dr Chris Purvis -- Development Manager ISDN Communications Ltd, The Stable Block, Ronans, Chavey Down Road Winkfield Row, Berkshire. RG42 6LY ENGLAND Phone: +44 1344 899 007 Fax: +44 1344 899 001