This came about as I didn’t have access to 448 cores of JVM Azul goodness. The pretotype used 20 over-clocked Raspberry Pi units and an old HP Superdome picked up from eBay (I needed something beefy for the database and it was cheaper than RDS). After a bit of light surgery with a sharp scalpel, aka Spring Remoting with Hessian, it was time to awaken Frankenstein’s monster. The results are very promising though there are still a few kinks to be ironed out, with the Solr 1.4 index being one and the shared database schema between microservices being a glaring architectural impurity.
As for the next iteration, well “I’ve decided to take my work back underground to stop it falling into the wrong hands”.