System for locating and anticipating data storage media failures

Abstract

Method and apparatus for use in a computer having peripheral data storage, to monitor data transfers to detect defective or marginally defective storage areas. The method and apparatus are based on a memory-resident program component and a transient program component which cooperate. The resident component is in the computer's interrupt chain, to be invoked during both the initiation and conclusion of storage operations. When a data access is called, a timer (e.g., the system timer) provides interrupts at a suitably high rate, and informs a second component of the operation. At the conclusion of the operation, error indications are logged into a log file. The second component accumulates interrupts during the operation, and monitors one or more memory addresses. For each monitored address, when its contents first change, the second component stores the current interrupt accumulation. Following the operation, the first component applies algorithms to the accumulations and decides thereby whether to log an error indicator. A third component places the first component in the storage call chain in one of two positions, including a preferred position providing raw status rather than processed status with some information removed. An analysis program retrieves the log entries, clears the log for future entries, adds the entries to a history of the storage, and applies algorithms to the history to determine whether storage areas are defective. It also takes remedial action such as marking a defective storage area bad and relocating the data to a usable area.

Dependencies are inferred from the claim text.