Process parameters:

    initial_load [neighbors...]

Communications:
     - two channels per host: normal and low_latency

How shall we manage link failures ?

Process model (?)

     while (there is something to do) {
        compute some task;
        get received tasks;
        compute load balancing;
        send tasks to neighbors;
     }

    * Open Questions :
     - definition of load on heterogeneous hosts?
     - how to detect convergence?
       : No need to detect convergence.  Computation stops when there
         nothing more to do.

     - how to manage link failures?

     - shall we retrieve pending tasks?
       : Ideally, why shall.  How?  By using some acknowledgment?