Re: Proposed text on reliability in the web services architecture

Miles Sabin wrote:
> Peter Furniss wrote,
> 
>>The cure, as Miles says, is to get the assurance of processing from
>>the processor, not from the intermediary.
> 
> [...]
> The problem is that arranging for those backend systems to signal a 
> failure back to the gateway, so the later can provide a nack back as 
> part of a WS reliable messaging protocol, might be difficult or 
> impossible. So if the gateway sends an ack back to the sender it could 
> be misreporting a failure as a success ... hence the connection with 
> byzantine failures.

I don't have much else to add here (it was Miles that educated me on 
this stuff me a few years ago anyway), other than the fact that 
gatewaying WS based RM to legacy is a very real issue, particulary 
having to mesh with legacy systems that are batch driven, have 
service windows or require the gateway to hold state in order to 
reconcille inbound and outbound messages because the interface to 
the legacy system 'can't' be changed to roundtrip information like 
message or correlation ids through (btw I'm not really thinking 
about banking systems). All this to think about as well as protocols 
like HTTP and SMTP. In any case I hope a model that can be used for 
WS RM falls out of this discussion.

Bill de hÓra

Received on Wednesday, 22 January 2003 05:13:00 UTC