Master worker model is a very popular paradigm for many parallel applications. The master generates tasks and workers consume them. In CometCloud tuple space paradigm is used for communication between master and workers, and it naturally supports BOT solution using the master worker model. The master inserts task tuples into the space and collects result tuples, and the workers extract task tuples from the space and insert result tuples. See Figure 1. However, CometCloud is not limited only to BOT solution. Master and workers can communicate with each other by inserting/getting communication tuples to/from the space.
The tuple space paradigm, made popular by Linda [1], includes: asynchronous communication that decouples senders and receivers in space and time; an associative multicast medium through which multiple receivers can read a tuple written by a single sender using pattern-matching mechanisms instead of names and locations; and a small set of operators (write, read, and remove) providing a simple and uniform interface to the tuple space.
It provides the following primitives:
- •out: an operation that inserts tuple into the space
- •in: an operation that removes a tuple matching a specified template from the space and returns it.
- •rd: an operation that returns a tuple matching specified template from the space. This operation performs exactly like the in operation except that the tuple is not removed from the space.
References
[1] David Gelernter. Generative communication in linda. ACM Transactions on Programming Languages and Systems, 7(1):80-112, 1985.