在 SVR4/MP 上 高 可 利 用 度 系 統 的 設 計 與 實 作

Abstract

近年來,需要大量計算時間的程式愈來愈多。可是一個系統難免會因為不 可預料之原因而發生錯誤而當機，導致時間和計算資源的嚴重浪費。一個 高可利用度系統的功能，在於能容忍不同的機能故障，而能持續的使用。 此篇論文中，我們在 UNIX SVR4/MP 作業系統上，設計以 及製作一高可 利用度伺服器，來有效地偵測並解決伺服器機能故障。此外，我們並提供 行程的檢查點與重新起始之功能。以便藉由極少量額外的硬體和軟體，達 到資料與行程的高可利用度。 In recent years, more and more programs need long execution time. But a system may become unavailable because of faults and which leads to the waste of time and computing resources. The functionality of a high availability system is to tolerate such faults and keep the usage of the whole system. In this thesis, we design and implement a high availability server, which can detect and resolve the server failure effectively, on UNIX SVR4/ MP perating system. Besides, we implement the checkpoint/ restart facility as user level commands. Thus, we can archive both data and process availability by using few extra hardware and software overhead.