Hilarie,
Note that John does NOT want the data to go from the place
where it is adapted back to the proxy/dispatcher. Thus, this is a
proxying case, not a callout server case for which OCP was designed.
If John uses OCP to talk between two processors/proxies, then some
default OCP Core assumptions will not work and will need to be
overwritten in OCP profile.
I agree that OCP can be used to send tracing and billing info
from one OCP agent to another, and that using OCP for such info makes
most sense if OCP is also used for application data.
Alex.
On Thu, 8 Jan 2004, The Purple Streak, Hilarie Orman wrote:
The tracing and billing informatio is slightly different. You say
that it might go back to the OPES processor(?) (through the
load-balancer?) or to some other location. This non-specifity of
destination is outside the OPES protocol, and you'd probably do best
to just have the callout servers open a direct connection to the
billing server. However, you'll need to think about whether or not
the OCP message headers give you enough information to correlate the
billing with the connection that triggered the callout server.
If you simply attach the billing data to the adapted stream and let
it flow back to the OPES processor, you'll need to write code that
recognizes and detaches that data and sends it to your billing
server. You could use OCP to communicate with the billing server, I
suppose, but there are probably any number of other protocols that
would be just as good.
Hilarie