I. OGG overview
OGG Full name Oracle Golden Gate.
History:
Established in 1995 in San Francisco, California, Golden Gate is named after the Golden Gate Bridge known as San Francisco. Two founders Eric Fish and Todd Davidson were originally designed to design a fault-tolerant system for tandem computer, which was used by banks to send transaction data from ATM networks to IBM mainframes due to the robustness of Golden Gate and the excellent data replication capabilities. Subsequently widely used in the financial industry and requires data replication efficient, robust industry. The company was acquired by Oracle in September 2009, before which Oracle and Golden Gate have more than 10 years of partnership. After Oracle's acquisition of Golden Gate, Oracle's original data guard complements each other, providing a solution for the enterprise to provide cross-platform real-time synchronization.
Golden Gate Software Classification Description:
(1) Golden Gate: Includes Goldengate Capture, deliver and active data Guard, XStream on Oracle DB, and supports a wide range of Open Data systems, which is our usual concept of OGG.
(2) Golden Gate Director: In fact, in the download, you can find no Director of the category, replaced by a Golden Gate Management pack package, in fact, this is the previous Ogg Director. Management Pack is a browser-based Administration toolkit with limited use of WebLogic SE, so configuring MP requires installing WebLogic first.
(3) Golden Gate Veridata: cross-platform comparison of replicated source and target libraries, providing data alignment and calibration capabilities for the source and target sides of Golden Gate.
In addition, there are also less common:
(1) Golden Gate for Mainframe: Includes Goldengate Capture, deliver for HP Non-stop, IBM DB2 on
Mainframe, plus syncfile.
(2) Golden Gate Adapters: interface packages for Java, Flat File, Siebel, and so on.
II. the OGG framework
supported heterogeneous platforms:
supported schema forms:
the functions implemented:
Iii. the OGG process in Oracle OGG 12c, the following components are available:
(1) Extract
(2) Data pump
(3) Replicat
(4) Trails or extract files
(5) Checkpoints
(6) Manager
(7) Collector
1. Extract Process
The
Extract process is responsible for data pump work. The Extract can be paired with the following two modes:
(1) Initial Load:
During the initialization of the load process, the Extract process extracts the current data directly from the source object.
(2) Change synchronization:
to ensure that the source data and other objects remain consistent with the data, the extract process captures the changes in DML and DDL since the source object was initialized for synchronization.
If rollback is received, extract clears all records before the transaction.
If you receive Commit,extract, the transaction information that was previously stored is persisted, the transaction information is written to the trail file on disk, and then the queue is routed to the target library.
2.Data Pump ProcessThe
Data pump process is a secondary extract Group on the Ogg Source library.
3. Replicat processThe
Replicate process runs on the target system, which reads the files from the target trail and resolves them to DML or DDL statements, and then executes on the target database.
The replicate can be configured in the following mode:
(1) Initial loads:
(2) Change synchronization
4. Trails FileIn
order to support continuous extraction and replication data changes, OGG crawls these changes to the data temporarily stored in a serialized disk file, this file is called Trail.
Trail files can be placed on the source or target system, and can even be placed on a temporary transit system, which depends primarily on the configuration of GG, if stored on the local system, called extract Trail or local trail, if stored in remote The system is called Remote Trail. Create and maintain trail files:
By default, trails is stored in the Dirdat subdirectory of the GG installation directory.
by default, each trails file is 10M, and a new trail file must be created constantly to keep the process running properly. This process is automatically implemented by Ogg and does not affect the operation of OGG.
for each newly generated file, the filename inherits the first 2 letters of the old trail file, and then adds 6 digits to make the uniqueness constraint. Numbers from 000000 to 999999. When the number reaches 999999, it continues to start at 000000. The previous trail file is overwritten.
add: (Extract Files)The
Extract file can be a document or be configured as multiple files. Multiple files can be configured to avoid a limit on the maximum size of a single file that touches an operation.
Extract file and trail files are similar, but checkpoints information is not stored in Extract file. Extract file is created automatically at run time.
5. Checkpoints mechanismreplicate will record the location of the transaction in Checkpoint table and checkpoint file. We can also not use Checkpoint table. 6. Manager ProcessThe Manager is the Golden Gate control process and the manager must run at all Golden Gate ends and run before the extract or Replicat process is started. During the extract or replicat process, it is necessary to keep the manager process running so that the associated resources can be managed. The Manager process has the following features:
(1) Start the Golden Gate process
(2) Initiate other related processes
(3) port for maintenance process
(4) Managing Trail Processes
(5) Create reports for Event,error and threshold. 7. Collector Process
a background process for target. The
Collector process does the following work:
(1) The request of the source-side extract process is sent to the manager process, the available ports are scanned and bound, and the port information is returned to the manager process, which is then assigned to the requested extract process.
(2) The Collector process receives the changed data sent by the source library extract process and writes the data to the target system trail file.
If
the srouce end pump process does not come up,
collector will lock the trail file. Then I need to confirm which pump process has a problem,View report
pump#, check to find collector, kill-9 kill. Additional notes:
each extract and replicat process requires approximately 25-55m of memory.
Ogg overview, architecture, Process
