Load balancing control method for a loosely coupled multi-processor
system and a device for realizing same
    1.
    发明授权
    Load balancing control method for a loosely coupled multi-processor system and a device for realizing same 失效
    用于松散耦合的多处理器系统的负载平衡控制方法和用于实现其的设备

    公开(公告)号:US4748558A

    公开(公告)日:1988-05-31

    申请号:US858791

    申请日:1986-05-01

    IPC分类号: G06F15/16 G06F9/50 G06F15/177

    CPC分类号: G06F9/505

    摘要: A loosely coupled multi-processor system including a plurality of terminals and corresponding processors. When a particular terminal tries to start a program, the corresponding processor sets the largest number of the terminals, which are serviceable, on the basis of the load status of the processors which are permanently watched by the OS (operating system) in a main memory. The processor compares this largest number with the number of terminals, which are in service at that time and when the number of the terminals, which are in service, is smaller than the largest number, processing by the terminal issuing a service demand begins. Alternatively, when the number of the terminals, which are in service, is equal to or greater than the largest number, the processor cannot treat the service demand from the terminal and reports this to a global processor. The global processor examines the load status indicator contained therein which shows the load status of each of the system processors; selects the processor having the lightest load status; and issues an order to treat the service demand from the requesting terminal to the selected processor having the lightest load. The processor, which has received the order, performs the processing of the service demand from the requesting terminal, based on the instruction from the global processor.

    摘要翻译: 松散耦合的多处理器系统包括多个终端和相应的处理器。 当特定终端尝试启动程序时,相应的处理器基于由主存储器中的OS(操作系统)永久监视的处理器的负载状态来设置可维护的终端的最大数量 。 处理器将此最大数量与当时正在使用的终端数量进行比较,并且当服务中的终端的数量小于最大数量时,由终端发出服务需求的处理开始。 或者,当正在使用的终端的数量等于或大于最大数量时,处理器不能处理来自终端的服务需求并将其报告给全局处理器。 全局处理器检查其中包含的负载状态指示器,其中显示每个系统处理器的负载状态; 选择具有最轻负载状态的处理器; 并且发出将请求终端的服务需求视为具有最轻负载的所选处理器的命令。 已经接收到订单的处理器基于来自全局处理器的指令执行来自请求终端的服务需求的处理。