Re: comments on draft-johnston-sipping-batch-notify-00

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Jonathan,

Thanks for your comments - see my responses below.

Thanks,
Alan


Jonathan Rosenberg wrote:
Alan,

Thanks for putting this together. First some questions:

* is this meant to do a one-time sync? or provide ongoing synchronization between two event servers? I think your primary use case is syncing state between a primary and backup, in which case it would seem to me you'd need this on an ongoing basis, no?

We have found use cases for both. In some cases, just a snapshot is needed, with some other mechanism being used for ongoing synchronization.

In other cases, perhaps the more common cases, the subscription will be continued. As a result, since there will be less state to transfer, the subscription will become a normal one, and probably not use these batch methods.


* Assuming the primary use case is syncing state between a primary and backup, these batch notifications don't seem sufficient to make that work. In the case of presence, you really need the input states to the compositor - whatever they are - in order to allow the backup to continue to compute the presence state.

Agreed - for presence, it would be the raw event state rather than any output from an event state compositor.

Similarly, in the case of conferencing state, the conference itself would need to move to the backup in order for the backup to continue to generate event state. If you've moved the conference over anyway, the backup has the state it needs to generate the event notifications. So I don't see what is served by the batch notifications.

I agree that the conference state use case is perhaps contrived - it isn't a real one that we've used it for. I need to think about it some more. However, I suspect it could be useful for syncing conference notification servers (as opposed to the actual focus doing the mixing and signaling).



* is the assumption that the subscription is to ALL state? Or do you intend that the body of the subscribe could contain, for example, a list of the URI for which state is desired?

This mechanism should work with filters, so the subscribe could include a filter body which would change it from all state to filtered state.



And a comment:

* the example shows the server terminating the subscription when it is 'done'. However, the state may have changed since it was sent to the backup. Seems you have the 'catch the running train problem' of database state sync and you'd need to continuously keep the subscription going.

Right - I think this is the more common case. In some cases, a different mechanism (such as publication) might be used to share ongoing state after the batch sync is complete. Does that make any sense?




Thanks,
Jonathan R.
--
Jonathan D. Rosenberg, Ph.D.                   111 Wood Avenue South
Cisco Fellow                                   Iselin, NJ 08830
Cisco, Voice Technology Group
jdrosen@xxxxxxxxx
http://www.jdrosen.net                         PHONE: (408) 902-3084
http://www.cisco.com

_______________________________________________
Sipping mailing list  https://www.ietf.org/mailman/listinfo/sipping
This list is for NEW development of the application of SIP
Use sip-implementors@xxxxxxxxxxxxxxx for questions on current sip
Use sip@xxxxxxxx for new developments of core SIP



_______________________________________________
Sipping mailing list  https://www.ietf.org/mailman/listinfo/sipping
This list is for NEW development of the application of SIP
Use sip-implementors@xxxxxxxxxxxxxxx for questions on current sip
Use sip@xxxxxxxx for new developments of core SIP

[Index of Archives]     [IETF Announce]     [IETF Discussion]     [Linux SCSI]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Big List of Linux Books]

  Powered by Linux