New and changed functions

PH53481

PTF UI95302

PH45481

PTF UI83586

PH45836

PTF UI83586

PH45836

PTF UI83586

PH45836

PTF UI83586

PH36767

PTF UI75665

When you are using IBM Db2 for z/OS version 12 and above and function level 500 or above SSID, PH46656 implements a user ID authorization check for HALOAD through DSNUTILB security validation. Applying the Db2 for z/OS APAR PH45948 is mandatory if authorization is managed by an SAF product, including RACF.

PH46656

(Replaces PH19777)

To use the Analytics Accelerator V7.1.2 interface, you must concatenate the Accelerator Loader product load library in the STEPLIB of the Db2 allied WLM environment that runs the SYSPROC.ACCEL_LOAD_TABLES stored procedure. See Using Analytics Accelerator V7.1.2 through V7.1.6.

• Db2 for z/OS data access methods
• Configuring Db2 Direct
• Creating virtual tables for RDBMS data sources

• Querying information in the IBM Application Discovery and Delivery Intelligence (ADDI) dictionary. See Configuring access to ADDI and Creating virtual tables for VSAM and sequential access using ADDI.
• Querying information in the IBM Rational Asset Analyzer (RAA) dictionary. See Configuring access to RAA, Creating virtual tables for VSAM and sequential access using RAA and Metadata Discovery preferences.

Delimited data can now be used with virtual tables. See Configuring delimited data support.

• Configuring rules and authentication for Big SQL
• Configuring rules and authentication for dashDB
• Configuring rules and authentication for LUW databases
• Configuring rules and authentication for Microsoft SQL Server
• Configuring rules and authentication for Oracle DRDA
• Configuring rules and authentication for QMF DRDA Server

You can specify a generation data group base name when defining a virtual table, which will cause all active generations of the group to be read. Using a virtual table rule, you can also read only a subset of a generation data group. See Creating virtual tables for sequential data and Configuring generation data set retrieval.

You must APF-authorize the hlq.SHLVRPC library. See APF-authorizing the load libraries (required).

In the Accelerator Loader studio, when generating JCL to load the accelerator, the option Enable Loader Parallelism now defaults to enabled. The default degree of parallelism is set to 4. See Generating JCL.

• When you load only the accelerator, a discard data set is now supported when running a load with a SYSREC data set. The DISCARDS keyword can also be specified to force Accelerator Loader to fail once a specified number of records are discarded.
• The IGNORE keyword of the Db2 LOAD utility is now supported. The IGNORE keyword controls how different types of discards are handled by Accelerator Loader. Ignored discards are not written to the discard data set and do not count towards the discard limit.
• A data conversion error will now cause Accelerator Loader to fail unless a discard data set is provided or IGNORE(CONV) is specified in the LOAD control cards. Previously, the record that caused the violation would be discarded and processing would continue.
• Accelerator Loader will now end with RC=0 even when records are discarded, provided the associated discard reasons are being IGNOREd. Previously, any discard would cause Accelerator Loader to end with RC=4. Since loads from an Accelerator Loader server data source do not support a discard data set, these loads will complete with RC=4 any time there are discards regardless of IGNORE settings.

When loading from an external file to the accelerator and Db2 (Dual load), if Accelerator Loader cannot determine the status of an accelerator from the ACCEL_CONTROL_ACCELERATOR stored procedure, the accelerator is considered unreachable and will be treated as offline. Previously, an unreachable accelerator would have caused the load job to fail.

When performing Consistent and Image Copy loads, Accelerator Loader performs additional row data validation prior to applying redo log records.

Target Unicode accelerator tables can be loaded from an Accelerator Loader server data source containing EBCDIC data. You can load EBCDIC data stored on the mainframe into target tables defined as CCSID Unicode. One of the primary use cases for this feature is compatibility between tables loaded from EBCDIC data and existing tables populated by other means. In particular, the accelerator does not support joins between Unicode and EBCDIC tables. Options CCSID and Enable Unicode Column Expansion in the Generate JCL to Load Accelerator wizard in the Accelerator Loader studio are provided for use when generating Accelerator Loader server load jobs.

In the Accelerator Loader studio, when creating virtual tables for CA IDMS data, the database name can be edited. When doing discovery, the studio gets back the first database name that is found relating to a record's area name; however, it is possible for records to be defined in multiple databases for the same schema/subschema combination. With the database name being editable in the virtual table map editor for CA IDMS data in the Accelerator Loader studio, the user can modify the map to point to a different database.

Accelerator Loader can now process SMF data in zEDC-compressed log streams. The Accelerator Loader server automatically detects when SMF log stream data has been compressed, and calls zEDC services to inflate the data in the log stream buffer before processing the SMF records.

When the primary Db2 subsystem to which Accelerator Loader is connected goes down, the Accelerator Loader started task now remains active and will automatically attach to another member of the data sharing group on the same LPAR, if applicable. Previously, this condition caused the started task to terminate. See Considerations for Db2 data sharing environments for more information.

MODIFY <started_task_name>,--REFRESH DB2

When loading from an external file to the accelerator and Db2, Accelerator Loader can issue a user-supplied return code when Db2 LOAD discards rows that Accelerator Loader has already delivered to the accelerator. A new Accelerator Loader started task option, RC_WHEN_DB2_DISCARDS, is provided for defining the return code. This new option does not change the return code when a row is discarded from both Db2 and the accelerator.

Accelerator Loader now offloads more CPU cycles to the IBM zSystems Integrated Information Processor (zIIP). This additional zIIP exploitation is available only when External load does not sort the data. External load sorts data only when loading a range-partitioned or index-partitioned object from a single SYSREC data set (that is, when partition level SYSREC data sets are not used).

Accelerator Loader now provides reduced CPU usage in cases where External load must wait for Db2 LOAD to build indexes. This performance improvement can reduce CPU usage when running an IDAA_DUAL type External load to a Db2 table that has indexes. CPU reduction is limited to cases where Accelerator Loader does not sort the data. Data sorting occurs only when loading a range or index-partitioned table from a single SYSREC data set.

When loading from an external file to the accelerator and Db2, you can optionally stop the target table space before loading a table enabled for replication. The stop drains all claimers and ensures that no updates are made to the Db2 table while the accelerator is being loaded. This behavior is controlled by the options module parameter STOP the target table space before initiating the load.

Changes have been made to reduce External load's CPU usage when running in task mode. The change that has the most impact will only reduce CPU usage in cases where External load does not sort the data. Sorting of data is currently only required when loading a range-partitioned or index-partitioned table.

• AES 256-bit can now be used to encrypt the password when the driver is establishing a connection with the server.
• A new composite SMF virtual table rule replaces all existing SMF virtual table rules. The updated member hlq.SHLVXVTB(HLVSMFT1) contains all the functionality in the previous SMF virtual table rules HLVSMFT1, HLVSMFT2, HLVSMFT3, HLVSMFT4, and HLVSMFT5.
• MapReduce and parallelism support is now available for accessing native IMS OSAM files. IMS compression exit support is also included for all supported IMS Direct database types.
• SQL access to CA IDMS record and set information managed by CA IDMS central versions running on z/OS is now provided. Virtual tables are mapped to IDMS records and sets that can be joined using standard SQL to navigate IDMS information. MapReduce processing is supported to improve performance of large data extracts of IDMS information.
• The ability to access IBM dashDB data sources via DRDA Virtual Request Facility (VRF) using standard SQL processing is provided. DRDA VRF is a feature that is designed to access data sources which provide the DRDA protocol.
• IBM Big SQL data sources are now uniquely identified by TYPE(BIGSQL) in the DEFINE DATABASE statement.
• In the Accelerator Loader studio, when virtualizing Db2/DRDA data sources, the user can select multiple Db2/DRDA tables and run the new wizard once to generate the required virtual tables. Previously, the user was required to run the Virtual Table creation wizard multiple times for each Db2/DRDA source table.
• The user can now set the data and index buffers for VSAM files. Two new Accelerator Loader server parameters, SQLENGVSAMDATABUFF and SQLENGVSAMINDEXBUFF, have been introduced to control these settings. Previously, the values were hardcoded.
• Virtual Parallel Data (VPD) now supports Adabas files; VSAM ESDS, KSDS, and RRDS files; and IAM files.
• IMS Direct now supports Fast Path data entry databases (DEDBs).
• A batch job with sample DRDARange and IMSRange commands, which are used to populate the Accelerator Loader server metadata repository, is now provided. The job is located in hlq.SHLVCNTL(HLVRANGE). Instructions for required edits to the job are provided in the member.
• MapReduce now supports Innovation Access Method (IAM) files. IAM is a VSAM optimization product distributed by Innovation Data Processing.
• A new set of SQL preferences has been added to the Accelerator Loader studio. These settings are related to SQL query generation, the SQL Results view, and SQL metadata retrieval. These new settings can improve the performance of metadata retrieval for Db2 and DRDA tables.
• A new Accelerator Loader server parameter has been introduced which specifies to automatically map all DECFLOAT columns defined in Accelerator Loader server virtual tables to DOUBLE at runtime. DECFLOAT is not a supported data type in the accelerator.

• The default size of the global variable file that is used by the Accelerator Loader server has been increased.
• The following server parameters have been added to the Product Parameters panel:
  • Enable support for SMF log streams and in-memory resources
  • Call the interface module for IAM
• The following changes have been made in the HLOIN00 template:
  • Parameters have been added for Virtual Parallel Data support.
  • Parameters have been added for enhanced MapReduce support.

A new method is available for terminating a Db2 utility for which DSNUTILB intercept processing is occurring or has occurred and performing the associated maintenance tasks related to DSNUTILB interception. In addition to using the Accelerator Loader maintenance utility HLOMAINT, you can now perform these tasks by specifying a new value, HLORESET, on the EXEC statement for the DSNUTILB program. Using this new method, you can terminate a stopped utility (if one exists), perform the associated maintenance tasks, and issue the new DSNUTILB request, all in one job. Previously, you had to submit an additional job to use the HLOMAINT utility separately to terminate a stopped utility and perform the associated maintenance tasks prior to running DSNUTILB.

Accelerator Loader now allows operation with only a single bootstrap data set (BSDS) defined for a Db2 subsystem. To define only a single BSDS for a subsystem, on the Db2 Subsystem Parameters panel , you can leave the Db2 Bootstrap DSN #02 field blank. Previously, a value in this field was required.

When performing a load from an external file (DUAL or ACCELERATOR ONLY profile), Accelerator Loader now supports SYSREC data sets in Db2 LOAD delimited file format.

When performing a load from an external file, Accelerator Loader can now convert string data from one character set to another when data is loaded from a SYSREC data set to the accelerator, Db2, or both. For example, you can now load data from an EBCDIC-encoded SYSREC data set to a Unicode Db2 table and the accelerator.

When loading the accelerator and Db2 from an external file, you can now control the action that Accelerator Loader takes when the Db2 LOAD utility discards records that have already been loaded to the accelerator. Using the Tools Customizer option Action when DB2 LOAD discards records loaded to the accelerator (Accelerator Loader started task options module parameter ACCEL_WHEN_DB2_DISCARDS), you can configure Accelerator Loader to leave the records in the accelerator and disable acceleration on the loaded table or to roll back the data loaded to the accelerator and leave the query acceleration status unchanged.

When loading data from a virtualized data source or remote DBMS, you can now load data to both the accelerator and Db2. Previously, when loading from these sources, you could load to the accelerator only. To use this feature, you must manually edit the JCL generated by the Accelerator Loader studio.

Accelerator Loader now supports the new Virtual Parallel Data (VPD) feature, which lets you group multiple simultaneous requests against the same data source and run them in parallel, while performing the input and output only once. For example, using VPD, users can load several SMF record types from a virtualized data source with only a single read of the data set. Previously, the data set would have been read once for each record type.

The default size of the global variable file that is used by the Accelerator Loader server has been increased.

A new keyword, CHECK_DATA, has been introduced to Accelerator Loader to enable you to specify if and when to check the integrity of Db2 for z/OS data pages. This keyword is available for Consistent and Image Copy loads. This feature has been introduced under the direction of IBM Support

Use Accelerator Loader to back up and recover data that resides only in the accelerator. This data can be in an accelerator-only table or an accelerator-shadow table that has been loaded to the accelerator only.

When performing an external load and BACKOUT YES is specified on the Db2 LOAD utility RESUME YES clause, Accelerator Loader recognizes when backout processing occurs and backs out the data sent to the accelerator, leaving the table with the same data it had before the load started. On an accelerator-only load, Accelerator Loader backs out the data sent to the accelerator when a data conversion error occurs. See Restrictions and considerations for adding data to a table (LOAD RESUME) in Restrictions and considerations for loading from an external file.

See Configuring support for code pages and DBCS for more information.

• Add the SYSAFF parameter to generated JCL for all non-Db2 tasks
• Create and use product staging libraries. Product staging libraries are a complete set of product data sets. Using staging libraries, you can retain customized modules when maintenance is applied to the product base libraries.
• Specify a UNIT value to use in some configuration jobs
• Include the Tools Customizer job card when generating the installation verification procedure (IVP) jobs
• Specify to use DSNUPROC (or another stored procedure) to run the utilities in the IVP jobs for each Db2 SSID
• Include STOGROUP and BUFFERPOOL in the IVP jobs
• Customize the Loader Policy for all SSIDs being customized

Accelerator Loader now supports accelerator groups, which allow users to load multiple accelerators by specifying a single accelerator group name. Users can specify either one group name or a list of accelerator names.

To access Oracle data, you can configure the Accelerator Loader server to access Oracle’s application server using a DRDA connection. Using Db2 LUW AESE or InfoSphere Federation Server is no longer required.

When performing an External load, you can load data to an accelerator-only table (AOT). VSAM objects do not exist in Db2 for AOTs; therefore, you cannot load to both Db2 and the accelerator (Dual load).

When performing an Image Copy load, you can take an image copy of a regular Db2 table and load the data into an accelerator-only table (AOT) with the same columns in the same order.

When loading data from non-Db2 and remote Db2 sources, when using the Accelerator Loader studio to generate the JCL, you can specify an accelerator-only table (AOT).

When performing an External load to only the accelerator, Accelerator Loader bypasses the Db2 LOAD utility when running a LOAD RESUME. This enhancement provides query access to the data while Accelerator Loader performs the load.

Accelerator Loader can load data from a remote system, enabling you to load remote data sources that are not directly accessible from the local system (not configured to the accelerator). To perform this type of load, an Accelerator Loader server must be running on the remote LPAR and the local server must be configured to communicate with that remote server.

Accelerator Loader supports IBM zSystems Data Compression (zEDC) to optimize cross-platform exchange of data by reducing network flow between the two servers when loading from a remote server.

Accelerator Loader supports Adabas as a selectable data source. To enable this feature, when customizing the product with Tools Customizer, specify the Adabas load library for the server to use to connect to the Adabas databases.

When performing a load from an external file, you can use a UNICODE SYSREC file to load data to a UNICODE TABLE.

When performing a load from an external file, you can load SYSREC data that is already in Db2 internal row format. The Db2 UNLOAD utility supports an option to unload the data from a table in FORMAT INTERNAL. This enhancement provides the benefits of reduced CPU consumption and elapsed time in both the UNLOAD and in Accelerator Loader jobs. Restrictions that the Db2 LOAD utility imposes when FORMAT INTERNAL is specified also apply to Accelerator Loader.