Java migrations…
Ralph Churchill has a good observation about Java App Server migrations, notably Migrating from Weblogic to JBoss.
“…and saved my company tens-of-thousands of dollars in licensing and (worthless) support costs.”
Oh the lure of big name, high-dollar Java application servers. They have a long list of features and are backed by big companies with promise of big support. Unfortunately many of the features don’t work as advertised, the support is difficult to access or non-existant, and in the long run the things turn into holes in your server you throw money in. I’ve had the same experience as Ralph, taking a Websphere installation (that I recommended and spearheaded) over to Resin (which I also recommended, after a couple years of Websphere pain and suffering). The difference was night and day, reliability shot up immediately and deployments went from all day affairs (maybe it will start this time) to 15 minute finger-snaps (done). Can’t recommend Resin highly enough, imho.
Anyway, what went wrong with the large Java app server players? Why can’t they deliver reliable software, actual support, all within reasonable dollar ranges (were normally talking order-of-magnitude dollar differences, and some are processor number based)? Small companies and open-source seem able to do this quite readily.