I had a very interesting debate recently about how long proprietary Message Oriented Middleware (MOM) would be able to hold out against the twin-pronged assault of REST (as the banner of common-sense integration) and open-standards implemented within open-source community. When I say ‘hold out’ I’m not talking of complete destruction, but more of accelerating commoditisation and loss of commercial significance to the vendor.
At that point opening the source-code and collaborating on unification within open-standards would possibly have more significance than bickering with competitors about who’s reliable QoS is better than the other as they eek out the remaining revenue from the tail-end of the customers who’ve not yet switched.
I base this assertion on a recent experience. I was involved in the implementation of an ODBC driver in the early 90’s just as ODBC was in the ascendancy. My role was the creation of a multi-protocol connectivity layer – one part of which was based on Windows-Sockets. There were a range of variations of this tool-kit and each was bound to a proprietary IP stack which had to be installed onto the host, and provided API’s through which Windows (16-bit) applications were able to drive the IP stack. I recall the various evaluations I performed, and the resounding conclusion was ‘there was no discernable difference’ !
Over time we saw the innovation around the IP stack melt into the operating system and become part of the baseline of expectation such that today, IP is as core to people’s perception of ‘the network’ as the very copper across which it’s electrons are transferred. So how is this relevant? Well the movement of IP from proprietary application-layer software, to uniform system-software, through to the range of firmware and optimised versions we see today, started with the recognition that is was commoditised and expected.
Granted the inertia to reach this tipping point in the early 90’s may have been driven by the all conquering Microsoft in it’s ascendancy, but today’s software industry is being overshadowed by the ascendancy of another heavy-weight, this time in the form of the open-source movement. I have no doubt that the existence of wire-level open-standards in MOM (i.e. AMQP) and the RESTian rebirth of HTTP in EAI, along with maturing open-source MOM implementations, will push the next layer of messaging innovation down into baseline of expectation in the near future.
It will be a measure of the vendors, to see which take confidence in their higher-order technologies and embrace the MOM community with the gift of their monopolistically-induced ‘experience’, versus those who cling to the last remaining units they can ship as a premium before losing their customer base.
Why is this good? Far better that we drive innovation higher up the food-chain to focus on solving real integration problems than burning time and ‘carbon’ worrying about which flavour of MOM is better! Let’s face it the Integration industry sorely needs some positive PR on the whole.
And now we’re hearing about putting application services into the network…