Re: I-D ACTION:draft-ietf-ldup-lcup-04.txt
John Merrells <[email protected]> Mon, 10 Mar 2003 20:58:54 -0800
| Newsgroups | gmane.ietf.ldup |
|---|---|
| Message-ID | <[email protected]> |
Jonghyuk Choi wrote: >Because the protocol does not explicitly exclude the so called 'no-history' >implementation, the server would have to return lcupReloadRequired >so often to achieve a reasonable degree of consistency. > Yep. >An optimization for this is that the client sends a lightweight >reload request to the server in response to lcupReloadRequired. >The server then sends the entries changed after the client cookie >as normal search entries and sends only DNs and UUIDs of unchanged >entries in order to reduce chattiness. The client can then consider >unarrived entries as deleted or scoped-out ones and delete them. > In this case the server is using some historical information to optimize the amount of information that needs to be sent back to the client. The more historical information stored the more you can optimize the amount of data that needs to be sent back. John