Cache Fusion

from Wikipedia, the free encyclopedia
Synchronization of two nodes in an Oracle database cluster bypassing the hard disk using Cache Fusion

Under cache fusion refers to the synchronization of two nodes in an Oracle database cluster bypassing the hard disk . In the Real Application Cluster (Oracle RAC), physically separate database buffers are used like a shared database buffer. This solves the problems of cache coherence control and the synchronization of the underlying shared disk architecture of the database system . Because a database block based on the logical shared everything architecture cannot be changed by many instances at the same time, there is no additional effort for buffer invalidation (cache coherence control, see e.g. Modified Shared Invalid Protocol ). Cache fusion is mainly implemented using two services ( Global Cache Service and Global Enqueue Service ). Other important concepts in Cache Fusion technology are past image and the global resource catalog .

Global Cache Service

The Global Cache Service (GCS) is a global service which manages the information on the database blocks used by all DBMS instances in the cluster and is responsible for the transfer of database blocks between the DBMS instances. This service enters all information about the database blocks used in the global resource catalog . In the RAC environment, various areas of the database must be protected by global locks (enqueues). This includes, for example, database catalogs, i.e. areas of the main memory that are intended for the internal administration of the database management system (DBMS) (e.g. library cache).

Global enqueue service

The Global Enqueue Service ( GES ) takes on the management of global resources in terms of blocking / exclusive access and the subsequent release as well as the detection and corresponding resolution of global deadlocks . The GCS and GES services are implemented as distributed services themselves . Each local DBMS instance has four additional processes that are used to synchronize simultaneous access and cache fusion.

Individual evidence

  1. a b c Larissa Janssen: High-performance database systems: Theory and Practice , Books on Demand GmbH, 2008, ISBN 9783833493263 , pp. 171–173.

Web link