ws-rx message
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [List Home]
Subject: mini-AI: problematic grouping flows
- From: Doug Davis <dug@us.ibm.com>
- To: ws-rx@lists.oasis-open.org
- Date: Mon, 27 Nov 2006 07:53:33 -0500
Per my mini-AI from the last call,
below are the sample messages that highlight the "grouping" problem
that the MSFT MC proposal has.
Using the current spec if the client
sends a series of requests to the server, each one may look something like
this:
<env>
<Header>
<wsa:To> myservice.com </wsa:To>
<wsa:ReplyTo> http://docs.oasis-open.org/ws-rx/wsrm/200608/anonymous?id=uuid1
</wsa:ReplyTo>
<wsa:MessageID> msgN </wsa:MessageID>
</Header>
<Body>
<someRequest>...</someRequest>
</Body>
</env>
Each response from the server may look something like this (before RM headers):
<env>
<Header>
<wsa:To> http://docs.oasis-open.org/ws-rx/wsrm/200608/anonymous?id=uuid1
</wsa:To>
<wsa:RelatesTo> msgN </wsa:RelatesTo>
</Header>
<Body>
<someResponse>...</someResponse>
</Body>
</env>
If the client uses the same RManonURI then the server can choose to group
all of the responses into the same RM sequence because it knows they're
all headed to the same endpoint. If we replaced the RManonURI with
wsa:Anonymous (as suggested by the MSFT proposal and the SeqID variant
of what's currently in the spec) then the responses will look like this:
<env>
<Header>
<wsa:To> http://www.w3.org/2005/08/addressing/anonymous </wsa:To>
<wsa:RelatesTo> msgN </wsa:RelatesTo>
</Header>
<Body>
<someResponse>...</someResponse>
</Body>
</env>
If the RM logic is not co-located with
the application logic (meaning its on a gateway or intermediary) then all
it has to examine to decide which RM sequence each message should go into
is this outgoing message. This means that it must either place each
message in its own sequence (which means no grouping at all) or all messages
sent to the anonymous endpoint will be in the same RM sequence - which
clearly won't work if there are multiple anon clients talking to the same
server at the same time. W/o some unique/identifying info (which
is just naturally there in the async cases) the RM processing model would
either need to be severely restricted or radically changed in the sync
case, w.r.t. RM sequence selection.
thanks
-Doug
__________________________________________________
STSM | Web Services Architect | IBM Software Group
(919) 254-6905 | IBM T/L 444-6906 | dug@us.ibm.com
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [List Home]