The Horus project is an ongoing effort. Although the initial version of Horus is nearing completion, we have yet to work out some important issues. For example, the current system can support several types of RPC mechanisms, and a great number of options exist for implementing state transfer (from a group to a joining member) and state merge. Selection of a preferred strategy for solving these problems and optimization of the corresponding mechanisms will occur in the coming months. Better support for flow control and resource management is another area of intensive activity.
We are also looking at running Horus on more advanced platforms. A specific target of our current work is stripped-down compute nodes over a high performance communication switch. We anticipate that this configuration of the system will expand our application domain to parallel computing, high performance I/O servers, multi-media, computer-supported collaborative work, etc. To enable these new types of applications, we are considering extending Horus to support real-time features and an object-oriented environment.
Another area of research is security and privacy. Our earlier work on Horus included a security architecture [12]. This work needs to be adjusted to fit in the layered structure of the current system. We are also pursuing research in anonymous communication, enabling message passing without explicit knowledge of the source and the destination addresses.