Résumé: A locking prctccol to coordinate access to a dietributed database and to maintnln system consistency throughout normal and abnormal conditions 1s presented in this paper. The protocol is robust in the face of [allures of any participating slte and in the face of netwcrk partit1oning. The proposed protocol supports the integratlon of virtually any locking discipline inclu11ng predicate lockir~ ~~thods. A cost and delay analysis cf the protocol as weIl as a proof o~ its cO:'rectness is included in this' worlc. The paper concludes with a proposal for an extension aimed at cptimizing operation of the protocol to adapt to highly skewed distributions of activity.

Los Angeles:
Langue: Anglais
Collation: 14 f. ;30 cm.

Mots clés:
Concurrency
crash recovery
Distributed database