RE: [Simple] Content indirection - Uploading large messages

[Cnaan Oded <[e-mail]>]

Sean,

Thanks for the prompt reply. 

You may be right that the details of how large payloads are sent from the UA
to the server (in case these are not the same entities and that there is a
network based server) are outside SIMPLE scope but it seems to me that the
first hurdle SIMPLE will face would be -interoperability. If you leave this
question open for proprietary solutions, applications from different vendors
will not be able to co-exist on the same network. If you look at the WG
charter, you will find that "The IETF has committed to producing an
interoperable standard for these services ". 

The charter also refers to RFC 2778 and CPIM which require defining how
messages are sent from UAs to servers, regardless of their size.

Moreover, if you look at 3GPP requirements from SIMPLE (e.g.,
draft-niemi-simple-im-wireless-reqs-00)  they require that "The content size
MUST NOT be limited by the Instant Messaging, or message session protocols"
- SIMPLE must be able to specify how this is done.

I believe SIMPLE WG should define at least a 'default' behavior for these
cases, or refer to an external standard, such as MM1.

Oded Cnaan


-----Original Message-----
From: Sean Olson [mailto:[e-mail]]
Sent: Tuesday, January 21, 2003 9:46 AM
To: 'Cnaan Oded'; [e-mail]
Subject: RE: [Simple] Content indirection - Uploading large messages


The draft intentionally does not address the how the content is placed
on the content server to begin with.
That is a separate problem that has many disparate solutions and does
not necessarily need to be standardized.
The interesting thing about standardizing the content-indirection
mechanism is that it involves two SIP entities
and directly impacts the SIP signaling. The transfer of content, either
from the UA to the server, or between UAs
is best handled outside of SIP ... the very point of the draft.

Some important things to note:

1) The "server" does not have to be distinct from the UA. They can be
one in the same.
2) The URL does not have to be an HTTP URL, though this will be probably
be one of the more common schemes used.
   For example, an LDAP URL might very be appropriate for certain types
of content
3) HTTP is ideal for conveying meta-data about the content it carries,
but that argument is best handled outside of SIMPLE
4) It was never envisoned that MESSAGE would be used to send these large
payloads -- the entire point of content 
   indirection is to offload this to a non-SIP carrier.
5) There is nothing mutually exclusive of content-indirection and MMS.
Nor is there anything that compels one to use
   MMS for this task in the draft. It's simply outside the scope of this
draft.
6) Is there something particularly non-concrete about the draft? ;-)
7) This mechanism is not tied to any particular architecture, though
some architectures may have more pressing need for
   it than others.

cheers,
sean

-----Original Message-----
From: [e-mail] [mailto:[e-mail]] On Behalf Of
Cnaan Oded
Sent: Monday, January 20, 2003 11:05 PM
To: '[e-mail]'
Subject: [Simple] Content indirection - Uploading large messages

The content-indirection draft (the latest I could find was
draft-ietf-sipping-content-indirect-02.txt) describes how messages or
notifications carry a URL in order to allow the UA to retrieve the
(usually large) content over other transports (e.g., HTTP). The draft
however does not specify how large content is uploaded from the UA to
the server.

Using HTTP per se is not sufficient for uploading messages over HTTP as
there is some meta-data that needs to be attached to the document such
as sender, recipient(s), message type etc. The MMS standard have already
solved these details in their MM1 protocol (over HTTP as well).

Note also that using the MESSAGE method is not a good practice as well
as it would generate huge load on the CSCFs (in mobile networks that
deploy IMS/MMD infrastructure), denying them from performing call
control tasks.

1. How should large messages be sent from the UA to the server? any
specific draft?
2. Does the SIMPLE WG intend to reuse the MM1 specifications or invent a
new one?
3. What's the status of this work within the SIMPLE WG? Are there any
concrete contributions?

Thanks,
Oded Cnaan
Comverse