The EMA has a sig that's trying to write a meta-standard for packages like
MSMQ and MQSeries. It's very laden with concepts already compiled into these
rpoduct exemplars: queues, queue management, but all in all, you have to
admit, they are handling a level of problems beyond the "Internet" paradigm of
eventual delivery. When's the last time you wanted to reprioriutize a remode
SMTP delivery queue, or undo a undelivered email?
What I can't see reflected as much in the media is the difficulty of
reverse-engineering applications to accept input from message queues -- the
wrappers. AM I missing something? were all these old CICS apps built qaround
some internal forms/query abstraction that's readily decoupled for
client-server use? or is there a lot of "cgi-bin'ing" going around?
InformationWeek last year gave MS MQ an editor's pick in a lengthy review. But
it's windows only. IBM's MQSeries got the nod from Infoworld 1/5/98 for
supporting >20 platforms, 5 network protocols, and many languages (including
C++, C, Cobol, Java, and PL/1). Up to 100MB per message, 1GB per queue.
This is all separate from transaction processing monitors, a whole nother
industry subset. But these packages together seem to be solving a lot of
problems for real businesses, cost O($50,000) to license and more to
understand, and hence make an attractive target for free software :-)
Seriously: there is something appealing at the core of message middleware
today which isn't being reflected in the academic discussions. Maybe it's the
interoperable message switch concept; maybe its the paranoid style of message
administration; maybe it's the hand-holding of a large consultant. But what
does it do that HTTP/NG/*TP shouldn't?
Can someone do to this industry what the Web did to hypertext?
Something from NetworkWorld
The Problem With MOM Over the past year, a similar trend
toward reliable business messaging middleware has emerged
out of the muddle of message-oriented middleware (MOM).
One of the perennial problems with MOM--beyond the
fragmented market created by myriad small vendors--is its
emphasis on the high-cost, highly customized end of the
market. The majority of MOM users are corporate IT
developers who use it to create expensive, homegrown,
mission-critical applications, typically in the
telecommunication or financial verticals.
To break out of this niche, current MOM market leader IBM (with its m
MQSeries) joined with future market leader Microsoft (with its recently
released MSMQ--formerly code-named "Falcon") and several other MOM
stakeholders (including Intel and Hewlett-Packard), to create the Business
Quality Messaging Special Interest Group (BQM-SIG) at the spring Electronic
Messaging Association conference (for more information see www.bqm.org). The
BQM-SIG is one of the factors pushing the ubiquity of transactional messaging
throughout client/server applications by the year 2000. (Or is traditional
messaging simply being pulled into existence by emerging business processes?)