General
Unrestricted Bulk Transactions
Oracle AI Database allows DML statements (INSERT, UPDATE, DELETE, and MERGE) to be executed in parallel by breaking the DML statements into mutually exclusive smaller tasks. Executing DML statements in parallel can make DSS queries, batched OLTP jobs, or any larger DML operations faster. However, parallel DML operations had a few transactional limitations.
This includes a limitation that restricted transactions with multiple per-table parallel DMLs. This means that once an object is modified by a parallel DML statement, that object cannot be read or modified by later statements of the same transaction. This enhancement removes this limitation, enabling users to run parallel DMLs, and any combination of statements like queries, serial DML, and parallel DML on the same object, within the same transaction.
For users, this simplifies and speeds up data loading and analytical processing by making full use of Oracle AI Database's parallel execution and parallel query capabilities.
ACFS Auto Resize Variable Threshold
ACFS auto resize now allows you to configure the threshold percentage for your file system automatic resize.
A more flexible threshold is now available for your file systems auto resize. Previously, the threshold was fixed to 10%. Now, you can customize to your specific use case needs.
ACFS Cross Version Replication
ACFS replication now allows for primary clusters to replicate to standby cluster on a previous or older release.
This feature will provide flexibility in replication configurations, providing ample time for upgrading and lifecycle maintenance.
ACFS Encryption Migration from OCR to OKV
ACFS Encryption now allows you to migrate from OCR to OKV.
This feature allows for a centralized point for key management using Oracle Key Vault.
ACFS Replication Password-less SSH Setup Tool
A new tool provides users the ability to configure SSH keys management for ACFS Replication.
Users can now avoid the repetitive, error-prone process of SSH keys management, setup, and configuration with this new tool. The tool makes the ACFS replication setup process more efficient and easier.
ACFS Replication Switchover
A new command, acfsutil repl switchover, provides a coordinated failover. However, if ACFS cannot establish contact the replication primary site, the command will fail.
Enhanced flexibility in ACFS replication management is now available with the addition of this new command.
ACFS SSH-less Replication
This feature provides an alternative transport choice for ACFS Replication which eliminates the need to maintain ssh-related host and user keys.
Users now have an alternative to ssh, including network data transfer, authentication between replication storage locations, encryption of the data stream, and a facility for executing remote commands.
ACFS Snapshots RMAN Sparse Backup and Restore
You can now back up and restore PDB snapshot copies on ACFS.
Backing up and restoring PDB snapshot copies on ACFS, provides the space-efficient storage that is inherent of ACFS Snapshots.
ACFS Sparse Backup and Restore of Snapshots
The acfsutil snap duplicate command can now generate a backup of an entire ACFS file systems and its snapshots, while preserving its sparseness.
You can now apply a full backup to another location while retaining the original sparseness. You can now replicate an entire ACFS file system and its snapshot tree with this new functionality.
ACFSutil plogconfig Log Files Wrapping Info
ACFSutil plogconfig offers you a way to manage persistent logging configuration settings. acfsutil plogconfig -q will now offer you additional information on whether the logs have wrapped or not. You can also get this information with acfsutil plogconfig -w, which will offer only this information and not all the comprehensive information offered by acfsutil plogconfig -q.
Further information regarding persistent logging is now available, hence enhancing the experience in the realm of diagnosability.
Automatic Parallel Direct Path Load Using SQL*Loader
The SQL*Loader client can automatically start a parallel direct path load for data without dividing the data into separate files and starting multiple SQL*Loader clients. This feature prevents fragmentation into many small data extents. The data doesn't need to be resident on the database server. Cloud users can employ this feature to load data in parallel without having to move data on to the cloud system if there is sufficient network bandwidth.
SQL*Loader can load data faster and easier into Oracle AI Database with automatic parallelism and more efficient data storage.
BIGFILE Default for SYSAUX, SYSTEM, and USER Tablespaces
Starting with Oracle AI Database 26ai, BIGFILE functionality is the default for SYSAUX, SYSTEM, and USER tablespaces.
A bigfile tablespace is a tablespace with a single, but large datafile. Traditional small file tablespaces, in contrast, typically contain multiple datafiles, but the files cannot be as large. Making SYSUAX, SYSTEM and USER tablespaces bigfile by default will benefit large databases by reducing the number of datafiles, thereby simplifying datafile, tablespace and overall global database management for users.
Bigfile Tablespace Shrink
This feature supplies the capability to reliably shrink a bigfile tablespace.
In earlier releases, organizations may have found that the datafile of a bigfile tablespace grew larger despite the actual used space being much smaller. This could happen after a user dropped segments/objects in the tablespace, but was not able to use datafile resize to recover the freed space due to the location of the data in the datafile.
By using Bigfile Tablespace Shrink, organizations can now reliably shrink a bigfile tablespace to close to the sum of the size of all objects in that tablespace, optimizing storage and reducing costs.
CEIL and FLOOR for DATE, TIMESTAMP, and INTERVAL Data Types
You can now pass DATE, TIMESTAMP, and INTERVAL values to the CEIL and FLOOR functions. These functions include an optional second argument to specify a rounding unit. You can also pass INTERVAL values to ROUND and TRUNC functions.
These functions make it easy to find the upper and lower bounds for date and time values for a specified unit.
CONSTRAINT_NOVALIDATE, a Data Pump Import TRANSFORM Parameter
The Oracle Data Pump TRANSFORM parameter option CONSTRAINT_NOVALIDATE enables you to set validation preferences. It has two options: Y or N. When set to Y, constraints are not validated during import.
Validating constraints during import that were valid on the source can be unnecessary and slow the migration process. Validation can be done after import.
Centralized Configuration Providers
Database clients can securely pull application configuration data from Azure or OCI Cloud. The store can contain data such as application connection descriptors and tuning parameters.
Central configuration makes application management and scaling easier. It fits well with architectures such as microservices and serverless deployments.
LONG_TO_LOB, a Data Pump import TRANSFORM parameter
The Oracle Data Pump TRANSFORM parameter now includes the LONG_TO_LOB option to migrate all LONG and LONG RAW data types.
Using this transform enables you to migrate the deprecated LONG and LONG RAW data types by transparently and automatically converting them to CLOB and BLOB data types. This parameter changes all LONG data types to CLOB and all LONG RAW data types to BLOB.
The syntax is TRANSFORM = LONG_TO_LOB:[Y|N]. The default is N.
Materialized Expression Columns
Expression columns, also known as virtual columns, are additional columns derived (computed) from existing columns. They can be persisted (materialized) on disk, complementing the existing default functionality of computing the results at runtime only.
Choosing between computation at runtime and computation at DML time for expression columns provides more flexibility in choosing the right approach for an application.
Materializing expression columns trades disk storage for the need to compute the same expression over and over again.
Oracle Data Pump Filters GoldenGate ACDR Columns from Tables
The ACDR feature of Oracle GoldenGate adds hidden columns to tables to resolve conflicts when the same row is updated by different databases using active replication. GoldenGate can also create a "tombstone table," which records interesting column values for deleted rows. Oracle Data Pump can exclude the hidden columns and the tombstone tables by setting a new import transform parameter, OMIT_ACDR_METADATA.
Oracle Data Pump enhances migration flexibility. It can migrate data from an Oracle GoldenGate ACDR (automatic conflict detection and resolution) environment to a non-ACDR environment by excluding the GoldenGate ACDR metadata during import.
PDB Snapshot Carousel ACFS Support
Oracle ACFS now supports PDB Snapshot Carousel, which allows you to maintain a library of PDB Snapshots.
Oracle AI Database files stored on Oracle ACFS file systems can now leverage PDB Snapshot Carousel in conjunction with ACFS snapshot technology.
SQL*Loader Supports SODA (Simple Oracle Document Access)
SQL*Loader now supports Simple Oracle Document Access (SODA). You can insert, append, and replace external documents into SODA collections in Oracle AI Database applications by using the SQL*Loader utility in both control file and express modes.
SQL*Loader support for Simple Oracle Document Access (SODA) makes it easier and faster to load schema-less JSON or XML-based application data into Oracle AI Database.
Sessionless Transactions
Managing a transaction requires the connection and session resources to be tied to the transaction throughout its lifecycle. Therefore, the session or connection can be released only after the transaction has ended. This often results in underutilization of sessions/connections. In Sessionless Transactions, after you start a transaction, you have the flexibility to suspend and resume the transaction during its lifecycle. The session or connection can be released back to the pool and can be reused by other transactions, therefore effectively being able to multiplex transactions and sessions/connections.
Sessionless Transactions provide ability for applications to suspend and resume transactions across sessions/connections (single instance or RAC) without the need for an external transaction manager, and without the application having to coordinate the commit and recovery protocols. The database manages transaction lifecycle, including commit and recovery. Application performance, and throughput, benefit from reduced commit latency since fewer client-server roundtrips are needed. Since external coordination is not required, using Sessionless Transactions results in vastly simplified mid-tier or app-tier infrastructure, and significantly decreases downtimes when compared with externally coordinating transactions (such as with XA).
XMLTYPE_STORAGE_CLAUSE, a Data Pump Import TRANSFORM Parameter
Transportable Binary XML simplifies the XML data storage and makes it easier to transport. It does not store the metadata used to encode or decode XML data in a central table.
XMLTYPE_STORAGE_CLAUSE now takes the options TRANSPORTABLE BINARY XML or BINARY XML. Oracle recommends that you use the TRANSPORTABLE BINARY XML XMLType, the new and recommended storage type for Release 26ai to store data in a self-contained binary format.
Use the BINARY XML (Non-Transportable) storage XMLType to store data in a post-parse, binary format designed specifically for XML data. Binary XML is compact, post-parse, XML schema-aware XML data. Binary XML is non-transportable and stores the metadata used to encode or decode XML data efficiently in a central table.
Data Pump can export and import data of type XMLType regardless of the source database XMLType storage format (object-relational, binary XML or CLOB). Oracle Data Pump exports and imports XML data as Transportable Binary XML so the the source and target databases can use different XMLType storage models for the data.