Young

On April 18th 2008, two Symphony DE engineers; Ajith and I, presented Symphony to researchers at MSRG (Middleware Systems Research Group) at the University of Toronto. This group works on various event management technologies using Pub/Sub message queuing paradigm. They have the objective of providing a more robust ESB (Enterprise Service Bus) to any SOA-based large-scale distributed systems.

The purpose of the talk was to gain the attention of the group by presenting the high-level architecture and the fault-tolerant characteristics of Symphony.

The MSRG researchers raised many interesting questions worthy of further consideration. Here are some of the questions;

• Is the Session Manager a bottleneck? Can you balance the workload through multiple Session Managers (SSM)? What's the complexity of synchronizing states among SSMs? How could multiple SSMs improve the scalability of the middleware?
• Can Service Instances share state or depend on each other?
• Why have multiple pairs of Service Instance Manager (SIM) and Service Instance (SI) in a compute host. How about having one SIM per host managing multiple SIs?

Through the talk, we verified that our technologies align well with their research direction and it opened up the possibility of forming an association to work on HPC middleware solutions together.

To learn more about MSRG and their on-going research projects, you can visit http://research.msrg.utoronto.ca/Padres/WebHome.
Comprehensive description of their key event management technologies is documented in the research paper, "A Policy Framework for Content-based Publish/Subscribe Middleware" , published in Middleware 2007 proceedings.