Issue No.04 - April (1986 vol.35)
D.J. Taylor , Department of Computer Science, University of Waterloo
A robust storage structure is intended to provide the ability to detect and possibly correct damage to the structure. One possible source of damage is the partial completion of an update operation, due to a "crash" of the program or system performing the update. Since adding redundancy to a structure increases the number of fields which must be changed, it is not clear whether adding redundancy will help or hinder crash recovery. This paper examines some of the general principles of using robust storage structures for crash recovery. It also describes a particular class of linked list structures which can be made arbitrarily robust, and which are all suitable for crash recovery.
storage structures, Crash recovery, global error correction, linked lists, local error correction, software fault tolerance
D.J. Taylor, C.-J.H. Seger, "Robust Storage Structures for Crash Recovery", IEEE Transactions on Computers, vol.35, no. 4, pp. 288-295, April 1986, doi:10.1109/TC.1986.1676761