Recovery and Coherency-Control Protocols for Fast Intersystem Page Transfer and Fine-Granularity Locking in a Shared Disks Transaction Environment

llbstract This paper proposes schemes for fast page transfer between transaction system Instances In a shared disks (SD) environment where all the sharing Instances can read and modify the same data Fast page transfer improves transaction response time and concurrency because one or more disk I/OS are avoided while transferring a page from a system which modified it to another system which needs it. The proposed methods work with the steal and no-force buffer management policies, and fine-granularity (e.g., record) locking For each of the page-transfer schemes, we present both recovery and coherency-control protocols Updates can be made to a page by several systems before the page is written to disk. Many subtleties Involved in correctly recovering such a page in the face of single system or complex-wide failures are also discussed. Assuming that each system maintains its own log, some methods require a merged log for restart recovery while others don’t Our proposals should also apply to dlstrihuted. recoverable file systems and distributed virtual memory in the SD environment, and to the currently oopular client-server object-oriented DBMS environments where the clients cache data.