This supports both Java JARs and .NET DLLS which are lightweight. You can also use language native data types where there is no need to data marshall with the cumbersome MWArray. This is also using HTTP as well making it no different thab Tomcat or JBOSS. Time 10:33 shows this exactly which really changes how to integrate Matlab with a Trading application. Mathworks addressed the big data with the ability to implement an algorithm or model against directly in your database. This can also be done in memory. It supports Hadoop cluster, app servers, JMS/MSMQ/IBMM data bus, database servers, etc. So you could add a Matlab algorithm to MapReduce statement. For a demo, set 21:22 using the ‘deploytool’. 27:22 demos how this can be done in Java. 28:11 demos how fast it is. 29:43 demos the connection to SAS database but it is mentioned you can connect into Oracle and SQL Server as well. 38:28 demos it through an |Apache web server like Tomcat. 35:00 demos a COM server application for Excel , .NET, and C#. This was introduced in Matlab 2012b.