Oracle 18c was released on 2018-02-16, still adhering to Oracle's Cloud first philosophy, 18c is now available on Cloud and Engineered Systems. Oracle 18c claims to be an autonomous database, which can reduce a lot of DBA work. Do many DBA staff worry about their work being affected? In fact, it is not particularly obvious on Oracle 18c, but this is a trend and signal in the future!
I checked some official documents and shared some new features with you. 18C continues the features of 12C's cloud database, and has changed the naming method of subsequent versions since 18C.
Before understanding the new features of 18C, it is necessary to simply have a basic understanding of 12C. On June 26, 2013, Oracle released Oralce 12C. Cloud computing design. 12C introduces new features of CDB and PDB, in the Multitenant Environment introduced by ORACLE 12C database (Note: Multi-tenancy technology (English: multi-tenancy technology) or multi-tenancy technology, is a software architecture technology, which is to explore and implement how to share the same system or program components in a multi-user environment, and still ensure data isolation between users), allowing a database container (CDB) to host multiple pluggable databases (PDB).
The full name of CDB is Container Database, which is translated into database container in Chinese, and the full name of PDB is Pluggable Database, which is a pluggable database. Prior to ORACLE 12C, instances and databases had a one-to-one or many-to-one relationship (RAC): that is, an instance could only be associated with one database, and the database could be loaded by multiple instances. An instance and a database cannot have a one-to-many relationship. When entering ORACLE 12C, the instance and the database can have a one-to-many relationship.
After you have a basic understanding of 12C features, you can learn about some new features added after 18C. Some of the new practical new features are listed below.
（一）Availability
- Oracle Data Guard Multi-Instance Redo Apply Supports Use of Block Change Tracking Files for RMAN Backups
Now, Multiple-Instance Redo Apply (also called MIRA) can also support BCT (Block Change Tracking) backup methods. This is a very effective incremental backup method for very large databases, where both the master and slave are RAC (real-time application clusters), and the backup occurs on the slave database. - Automatic Correction of Non-logged Blocks at a Data Guard Standby Database
Automatically corrects the bad blocks caused by Nologging in the standby database. Two new standby logging modes are added (mainly to speed up the loading data of the main library):
One is Standby Nologging for Data Availability, that is, the commit of the loading operation will be delayed until all the standbys have applied data.
SQL> ALTER DATABASE SET STANDBY NOLOGGING FOR DATA AVAILABILITY;
One is Standby Nologging for Load Performance, this mode is similar to the previous one, but when loading data, when network bottlenecks are encountered, data will not be sent first, which is The loading performance is guaranteed, but the data is lost, but the lost data will be obtained again from the primary.
SQL> ALTER DATABASE SET STANDBY NOLOGGING FOR LOAD PERFORMANCE; - Shadow Lost Write Protection
creates a shadow tablespaces (note: big file tablespace) to provide protection. (Note: At this point, you don't need ADG to provide additional lost write protection), write lost shadow protection can be enabled at the tablespace, database, and data file levels to proactively check in advance and prevent write loss. - Backups from non-CDBs are usable after migration to CDB
The original non-CDB can be migrated as a PDB to an existing CDB in this way. - The adaptive sequence of Scalable Sequences
is to reduce the competition under high-concurrency DML. By constructing discontinuous sequences, break up and reduce the split competition on similar indexes, etc., which is an enhancement from Oracle's optimization best practices. - User-Defined Sharding Method
supports user-defined sharding. This feature, which was present in the beta version of 12.2, was removed in the official release. Now, the release is out again. - Consistency Levels for Multi-Shard Queries
provides the MULTISHARD_QUERY_DATA_CONSISTENCY initialization parameter, which can be set before execution to avoid SCN synchronization during cross-shard query. - Manual termination of run-away queries
You can now manually kill a statement without disconnecting the session:
ALTER SYSTEM CANCEL SQL.
（二）Big Data and Data Warehousing
- Approximate Top-N Query Processing
Note: In 18c, APPROX_COUNT and APPROX_SUM are added to cooperate with APPROX_RANK. - LOB support with IMC, Big Data SQLLOB
objects also support in memory.
三）Database Overall
- Copying a PDB in an Oracle Data Guard Environment adds 2 new parameters to facilitate the creation of PDBs in an ADG environment.
One is STANDBY_PDB_SOURCE_FILE_DIRECTORY, which automatically finds the data file path of ADG (note, before 18c, if you insert a pdb into cdb in a standby environment, you need to manually copy the file to the OMF path of pdb) The other is STANDBY_PDB_SOURCE_FILE_DBLINK, which is convenient for remote Automatically find the ADG file path when cloning (Note: Before 18c, if it is a local clone, there is no need to copy the data file, but for a remote clone, it needs to be copied manually). - PDB Lockdown Profile Enhancements
can now create PDB lockdown profiles in application root and CDB root.
You can now create another pdb lockdown profile based on one pdb lockdown profile.
18c includes three default lockdown profiles: PRIVATE_DBAAS, SAAS, PUBLIC_DBAAS - Refreshable PDB Switchover PDB refresh
supports switchover. switchover is divided into two scenarios that are not planned by the planned kernel. Planned, can be cut back, mainly used to balance the load of the CDB. Unplanned, mainly used for the failure of the PDB master, without switching the entire CDB. - PDB Snapshot Carouselpdb's snapshot
backup carousel, which retains 8 copies by default, backed up every 24 hours.
ALTER PLUGGABLE DATABASE SNAPSHOT MODE EVERY 24 HOURS; - New Default Location of Oracle Database Password File
The password file path is already in ORACLE_BASE, not ORACLE_HOME, to remove changes to ORACLE_HOME. - Read-Only Oracle Home
can be installed in dbca or roohctl -enable to process read only oracle home. Run the orabasehome command to check whether the current Oracle Home is read-only. If the output of this command is the same as ORACLE_HOME, it means that Oracle Home is available. read and write. If the output is ORACLE_BASE/homes/HOME_NAME, it means Oracle Home is read-only. Set ORACLE_HOME to read-only, then change files will be created in $ORACLE_BASE, which is convenient for standardization, distribution sharing, rolling upgrade, etc. - Online Merging of Partitions and Subpartitions
supports online merging of partitions and enhances online maintainability. Note: The ONLINE keyword is required. - Concurrent SQL Execution with SQL Performance Analyzer
SPA can run in parallel (by default it is still serial), helping you complete SPA tests faster, this feature is useful when upgrading.
(4) Performance
- Automatic In-Memory
Automatic In Memory will evict the infrequently accessed IM columns from memory according to the Heat Map under the condition of tight memory usage. - Database In-Memory Support for External Tables
External tables support the IM feature. - There is a memoptimize pool area in Memoptimized Rowstore
SGA. The size is set by the MEMOPTIMIZE_POOL_SIZE parameter. When fast lookup is enabled, this memory area can be used for fast search. To enable fast lookup, you need to add keywords to the table creation statement: when querying based on the primary key, fast lookup can be used. Memoptimized Rowstore will greatly improve the high frequency query based on primary key in IoT.
（五）RAC and Grid
- ASM Database Cloning
ASM database cloning supports multi-tenant data. This feature provides a data-based native cloning method through ASM redundancy, which can replace storage-level-based cloning or replication synchronization - Converting Normal or High Redundancy Disk Groups to Flex Disk Groups without Restricted Mount
encourages the transition to flex diskgroup.
(6) Security - Integration of Active Directory Services with Oracle Database
and Microsoft AD. Before 18c, it was necessary to use Oracle Enterprise User Security (EUS) for interaction. Now, users and groups of AD and users and roles of Oracle can be mapped directly using centrally managed users (CMU).
The above are the main features involved in the version released by Oracle 18c. More new features need to be discovered and summarized in the process of practice. However, from the perspective of Oracle's development strategy, the later versions will become more and more Intelligence, catch the spring breeze of cloud computing and artificial intelligence, we will wait and see, looking forward to Oracle to bring us more surprises!