Portal de Periódicos da CAPES

The Benevolent Bandit Laboratory: a testbed for distributed algorithms

1989; Institute of Electrical and Electronics Engineers; Volume: 7; Issue: 2 Linguagem: Inglês

10.1109/49.17702

1558-0008

Robert E. Felderman, Eve M. Schooler, Leonard Kleinrock,

Parallel Computing and Optimization Techniques

The design, implementation, and use of a distributed processing environment on a network of IBM PCs running DOS is described. Temporarily unused PCs can be accessed by other users on the network to perform distributed computations. An owner of a PC need not be aware that the machine is being used during idle times; the machine is immediately returned when the owner begins to work again. Some degree of computation resiliency is provided in this unreliable environment; if a PC is part of a distributed algorithm and is reclaimed by its owner, the system finds a replacement node (if possible), resends the affected code to the processor, and restarts it. Thus, a distributed computation is able to proceed despite a set of transient processors. System performance, distributed applications, and fault tolerance are discussed. Performance improvements are demonstrated by applications like parallel merge sort and a distributed search solution to the eight puzzle. >