Toward Very Large, Efficient, Usable Network Structures
暂无分享,去创建一个
This chapter discusses some of the problems of programming networks, and of developing programming languages and good interactive software tools. Each particular problem, for which one develops an algorithm and a program, has its own structure. Ideally, one should build or reconfigure a computer network that embodies this structure as closely as possible. Occasionally that is possible, when the problem is important enough and the program runs long enough on sufficient data to justify the expense. In such cases, one may achieve a network structure that exhibits the anatomy of the problem, its nodes processing information flowing through it much as a human brain/nervous system processes information. More frequently, the same network will have to process a variety of programs, with a variety of different underlying structures. Therefore, one needs more general networks. An interesting, relatively simple network is one that contains a variety of different kinds of resources; each specialized for particular kinds of processes. Such a system is attractive because it also includes the conventional host 1-CPU computer that can be used for program development, editing, compiling, and the other traditional support processes.