I have been hearing a lot about Apache Thrift recently, though I understand hardly any about this. I realize it's a remote procedure call framework and abstracts calling functions across languages as well as on different machines. I have investigated MPI and located it's absurdly low-level. Would Thrift be considered a good greater level alternative to permit parallel computation to become carried out on the networked number of machines?
The solution is dependent in your performance needs. If you're searching for pure computational energy while using networked number of machines then thrift is less than ready.
Thrift features its own serialization to abstract away the kind conversion between languages and versions from the API. This is ideal for enterprise/client server systems which could go ahead and take performance hit to do these conversions for that benefits which come from permitting clients and servers in various languages. But also for a higher performance networked number of machines this might be useless as the nodes will most likely make use of the same language.
Even the asynchronous I/O is rather new and immature for the majority of the languages meaning using obstructing network I/O. This really is most likely not well suited for what for you to do.
I personally use thrift extensively also it solves lots of problems and also the community is rather active. Nonetheless its most likely not the best tool for the problem.