ORIGINALLY POSTED 17th August 2007

9654 Views on developerworks

Over at ‘zilla central Mark is discussing the many millions of dollars devoted to research and development products, and in particular how few of them actually make it out into the field as fully fledged products.

Back in the late 90’s IBM’s Research lab in Almaden were busy constructing just such an R&D project, codenamed ‘COMPASS’ – essentially a close to real-time clustering development environment aimed at using commodity parts as its hardware base. Since this was coming out of the storage research team, it was aimed in particular at exploring the possibility of building an enterprise clustered storage controller from said common parts. Around the same time a couple of Systems & Technology Group (STG) developers from our team in Hursley were on a placement working within the same group in Almaden. When they returned they brought back with them the idea of actually building a product using ‘COMPASS’ at its heart. Four years later and SVC 1.1.0 was GA’d, not only did this require a vast investment of resource, but the capital outlay to build a lab to test a heterogenous SAN virtualization appliance was huge.

Here we are, four years later on release 4.2.0 with about 10 major releases in between and a field population of over 3000 Clusters, yes clusters, not nodes (over 10,000 nodes have been shipped I believe). Anyway, many thanks go to the Almaden team – and keep them coming!

So some of them do actually break through and go on to create a new marketplace or bring a new technology to the enterprise.