ABSTRACT

This chapter describes the general architecture and purpose of GoldenGate, divided up as a description of various replication methods, followed by detailed internal architecture, including various processes, checkpoints, extract trails, groupings, internal change sequencing, and globalization. Database changes are divided up into two different categories: Data Manipulation Language that changes data in tables, and Data Definition Language that changes tables and other metadata objects. The Manager process starts and stops GoldenGate processing, any dynamic processes, assigns port number allocations, purging of trail files, as well as managing reporting. The Extract process captures changes from a source database and usually runs on the source database but can run on a target database as well. Master-to-master replication copies information from all servers to all other servers, where each is both source and target. The Collector process runs on the target database, writing changes from the source database into the trail files, to be consumed by the Replicat process on the target database.