gluepy: A Framework for Flexible Programming in Complex Grid Environments

Problem-solving frameworks in large-scale and wide-area environments must handle connectivity issues (NAT and firewalls), maintain scalability with respect to connection management, accommodate dynamic processes joining/leaving, and provide simple means to tolerate communication/node failures. We design and implement such a framework by minimally extending distributed object-oriented models for maximum generality and flexibility. In the framework, parallelism is expressed via asynchronous method invocations to allow a natural transition from sequential programs. To cope with asynchronous events such as dynamic joins and asynchronous method invocation returns, we introduce an implicit serialization semantics on objects to relieve programmers from explicit synchronization primitives while avoiding recursion deadlock problems. In our implementation, participating nodes automatically construct a TCP overlay so as to address connectivity and scalability issues. We have implemented our framework as a library for Python to allow rapid development of complex workflows and to maximally leverage the richness of its libraries. For evaluation, we show on over 900 cores across 9 clusters with complex networks (involving firewall and NATs) and process managements (involving SSH, torque, and SGE), how a simple branch-and-bound search application can be expressed simply and executed easily.