Good point on the draw back of any single wire protocol like IIOP/DCOM. But then also please see OMG specs for IIOP enabled firewall specs ( Orbix Wonderwall, and Visigenic's Gatekeeper are implementations of the same ).
One clarification, cant we have RMI/JRMP, RMI/IIOP, RMI/SOAP? If required the implementor decides which protocol to use beneath. Would that be a considered a bad design?
Regards,
Atul.
> -----Original Message-----
> From: Rickard �berg [mailto:[EMAIL PROTECTED]]
> Sent: Friday, March 17, 2000 1:14 PM
> To: [EMAIL PROTECTED]
> Subject: Multi-protocol vs Single-protocol - example scenario
>
>
> On multi-protocol vs single-protocol (i.e. RMI/IIOP)
>
> Consider this.
>
> Let's say we have a B2B scenario, where one company (A) wants
> to access
> a EJB-cluster of another company (B). Each company has a firewall. The
> firewalls only allows traffic on port 80. In order for A to be able to
> talk to B it must use a protocol that uses HTTP-tunneling. Let's say
> SOAP. By doing this A can access B, even though there are a number of
> firewalls in between. B, however, wants to access its EJB-cluster
> internally through some cool intranet web-application. They don't want
> to use SOAP, because it is not very fast. So, instead they want to use
> the protocol MyProt, which their cluster supports. MyProt is a very
> optimized protocol, and especially suited for cool intranet webapps.
>
> Result:
> A can access B's EJB-cluster through SOAP. B can access their own
> cluster through MyProt. Both are happy.
>
> If instead we had used RMI/IIOP only, the single-protocol alternative,
> we're already in trouble as the firewalls don't allow it. A is not
> happy. B, who accesses their own functionality through a web-app., can
> use RMI/IIOP, but their application passes a lot of XML-trees back and
> forth between the web-app and the EJB-server. IIOP doesn't handle that
> particularly well, so the result is a sluggish webapp. B is not happy.
>
> Result:
> Neither A nor B are happy with the solution. "EJB sucks" they think.
>
> Summary:
> Multi-protocol: A and B are happy.
> Single-protocol: A and B are unhappy.
>
> See. That was easy :-)
>
> This example was perhaps a bit exagerated, but definitely not
> unlikely.
> I wouldn't be surprised if many apps use this kind of setup
> in the (very
> near!) future. And if we then are restricted to one wire
> protocol, that
> doesn't work in many cases, that's bad for EJB. Which is bad for You.
>
> Any counter-arguments? I'm not an IIOP-expert (although I
> currently have
> the book "IIOP Complete" on my desk :-), so if you want to provide a
> solution to the above scenario using IIOP, please do.
>
> regards,
> Rickard
>
> --
> Rickard �berg
>
> @home: +46 13 177937
> Email: [EMAIL PROTECTED]
> http://www.dreambean.com
> Question reality
>
> ==============================================================
> =============
> To unsubscribe, send email to [EMAIL PROTECTED] and
> include in the body
> of the message "signoff EJB-INTEREST". For general help,
> send email to
> [EMAIL PROTECTED] and include in the body of the message "help".
>
