The Cluster Edition clusterware is integrated directly into Adaptive Server. No external clusterware is required to run Adaptive Server. Some of the components are new for the Cluster Edition; others are cluster-aware extensions of the existing Adaptive Server infrastructure. Figure 1-3 illustrates these components.
Figure 1-3: Key cluster-aware components of an instance
These are new, native cluster infrastructure components.
Cluster membership service – manages cluster membership and detects and handles instance failure.
Cluster interprocess communication (CIPC) – provides messaging services and an interconnection abstraction layer that allows the instances to communicate with each other via redundant pathways.
Cluster event service – supports a generic event-publishing and subscription mechanism for cluster-wide events.
These key components in the Adaptive Server DBMS layer have been extended to work in the Cluster Edition environment:
Buffer cache coherency – handles coherency issues related to the shared buffer cache and supports cache-to-cache transfer for allocation pages, index pages, data pages, object allocation map (OAM), and global allocation map (GAM) pages.
Cluster lock manager – supports distributed locking for coherency control across the cluster.
Cluster logging and recovery – handles logging from all instances, and fail over database recovery.
Cluster space and threshold – handles space and threshold management in distributed environment.
Object coherency – handles coherency issues related to sharing and transferring metadata and global variables. Object coherency must serialize updates to shared objects and make the latest changes available to all instances in the cluster.
Workload manager – an Adaptive Server module that provides application-level management of resource allocation, availability, and load distribution.