CDC Guide for Linux, UNIX, and Windows

10.2
- 10.2.2
- 10.2.1
- 10.0

Back Next

Oracle Data Sources

PowerExchange for Oracle provides two alternative methods of capturing change data from Oracle sources: PowerExchange Express CDC for Oracle
and PowerExchange Oracle CDC with LogMiner
. Both methods are delivered as part of the PowerExchange for Oracle CDC option but you can use only one of them in a PowerExchange instance with the same configuration files.

With both methods, Informatica strongly recommends that you use the PowerExchange Logger for Linux, UNIX, and Windows.

PowerExchange Express CDC for Oracle: PowerExchange Express CDC reads change data directly from Oracle active logs and from archived redo logs, including copies of the archived redo logs on a file system. PowerExchange Express CDC is more efficient and faster than PowerExchange Oracle CDC with LogMiner in many environments. Also, it avoids LogMiner reinitialization issues.
You can run PowerExchange Express CDC on the database server or on another supported 64-bit Linux, UNIX or Windows machine. At initialization, PowerExchange Express CDC stores the Oracle data dictionary in memory.

You must run Oracle in ARCHIVELOG mode with minimal global supplemental logging enabled.

PowerExchange Express CDC supports RAC and non-RAC environments, ASM, and Oracle Data Guard logical and physical standby databases.
PowerExchange Oracle CDC with LogMiner: PowerExchange Oracle CDC with LogMiner uses Oracle LogMiner continuous mining to read change data from Oracle active logs and from archived redo logs that reside at the archive destination to which they were originally written.
You must run Oracle in ARCHIVELOG mode with minimal global supplemental logging enabled. You must also periodically copy the Oracle online data dictionary to the archive log destination so that PowerExchange can determine restart points for change data extraction processing.

PowerExchange Oracle CDC with LogMiner supports RAC and non-RAC environments, ASM, and Oracle Data Guard logical standby databases.

You cannot use both PowerExchange Express CDC for Oracle and PowerExchange Oracle CDC with LogMiner in the same PowerExchange instance with the same dbmover.cfg and pwxccl.cfg configuration files.

The following table compares these PowerExchange Oracle CDC solutions:

Feature	PowerExchange Express CDC for Oracle	PowerExchange Oracle CDC with LogMiner
PowerExchange Logger for Linux, UNIX, and Windows use	Strongly recommended	Strongly recommended
Multithreaded processing	Yes	No
CDC processing speed	Faster	Slower
Oracle redo logs processing	Reads change data directly from active logs and from archived redo logs, including copies of the archived logs that were created outside of the Oracle archive process.	Uses Oracle LogMiner continuous mining to read change data from active logs and from archived redo logs that reside only at the archive destination to which they were originally written.
Oracle data dictionary use	Transparently stores the data dictionary in memory.	Requires that you create a copy of the data dictionary in the Oracle archived redo logs and perform periodic dumps to refresh the copy.
Capture from Oracle RAC and ASM environments	Yes	Yes
Capture from Oracle Data Guard environments	Can capture data from Data Guard logical and physical standby databases.	Can capture data from Data Guard logical standby databases only.
Capture from Oracle 12 c multitenant environments	Can capture data from a single pluggable database (PDB)	No
Capture from Amazon Elastic Compute Cloud (EC2) environments	Yes	Not certified in this type of environment.
Capture from Oracle Exadata machines	Yes	Yes
Capture from tables that use Oracle Exadata Hybrid Columnar Compression (EHCC)	Yes, except direct-path operations.	Yes, without restrictions.
Capture from objects that use Oracle Advanced Security Transparent Data Encryption (TDE)	Can capture data from encrypted tablespaces but not from encrypted columns.	Can capture data from encrypted tablespaces and columns.
Capture from objects that use Oracle Advanced Compression	Can capture conventional and direct-path DML operations from tables and table partitions and subpartitions that use Advanced Compression.	Can capture conventional and direct-path DML operations from tables and table partitions and subpartitions that use Advanced Compression.
Direct-path operations	Can capture direct-path operations except for tables that use EHCC.	Can capture direct-path inserts if the table or tablespace is set to the LOGGING and the load type is conventional path.
DDL operations	Does not capture CREATE TABLE…AS SELECT operations because the table cannot be registered for CDC. Tolerates ALTER TABLE ADD, ALTER TABLE ADD PARTITION, ALTER TABLE ADD CONSTRAINT, CREATE USER, ALTER USER, and DROP USER operations. The DDL change is not captured but CDC processing of other changes continues.	LogMiner maintains its own catalog of DB2 metadata so does not need to process most DDL changes. However, if a column with the NOT NULL attribute is dropped, an Oracle error might occur. If a CREATE TABLE…AS SELECT operation occurs, the table is not of interest for capture processing because it cannot be registered.
Oracle RESETLOGS events	Can capture data across a RESETLOGS boundary in the archive logs.	After a RESETLOGS event, the PowerExchange Logger fails to start. For a workaround, see Knowledge Base article KB425263.
EXCHANGE PARTITION operations	Does not capture the exchange operation or any rows it generates. Can capture subsequent DML changes on the table or partition that was the target of the exchange, if registered for CDC.	Does not capture the exchange operation or any rows it generates. Can capture subsequent DML changes on the table or partition that was the target of the exchange, if registered for CDC.
Source data loaded with the SQL*Loader utility	Can capture the data if the utility load type is conventional path and the load method is Insert, with the exception of data that is compressed with EHCC.	Can capture the data if the load type is conventional path and load method is Insert, Append, or Replace.
Capture from index-organized tables (IOTs)	Yes	Yes
Capture from materialized views	Can capture data from the master tables that underlie the views.	Can capture data from the master tables that underlie the views.
Capture from tables that use system partitioning or reference partitioning	Yes	No
Capture from tables in a sorted hash cluster	No	No
Capture from virtual columns with derived data	No	No
Capture from LOB columns	No	No