Features Introduced in Previous Versions
New Features in Version 6.2
Automatic Partitioning
When creating a partitioned table, the AUTOMATIC partitioning scheme randomly distributes rows evenly among the partitions.
AUTOMATIC is an alternative to the HASH partitioning scheme, used in earlier releases, in which rows are distributed evenly among the partitions according to a hash value.
For more information, see CREATE TABLE and Automatic Partitioning in the SQL Language Guide.
Query Result Caching
When a query is executed, the result is cached for a period. At the end of the period, the result is purged from the system. This feature can substantially reduce query time because Vector bypasses query execution and instead retrieves the result directly from the cache.
Query result caching will improve performance only when there are multiple requests of the same select query.
Typically, query results are reused if certain conditions are met. For more information, see Query Result Caching in the System Administrator Guide.
Spark Vector Connector 3.0
When you import data (e.g. via JDBC) using external tables and Spark, the staging table feature allows for preprocessing the data, e.g. data cleaning or schema mapping, on Spark side before loading it into Vector. This feature is supported only with Spark Vector Connector 3.0 in conjunction with Spark 3.0
For more information, see Load Data with Spark SQL through the Spark-Vector Connector in the Getting Started Guide.
UUID support for the ODBC driver
With the introduction of UUID support for the ODBC driver, automatic generation or import of UUID identifiers for inserting data can be done. UUID can be used as a primary key or as a partition key to ensure that data is spread evenly across nodes.
Workload Management
Enhancements to the Workload Management feature provide the ability to queue workloads and more finely control how workloads are executed.
For better control over queries, the database administrator can configure workload management queues for users, groups, and roles. Queues can then be configured to allow queries of certain users, such as a CEO, to be executed before other queries. System defined queues exist to allow certain types of queries, like short running queries or queries against system catalogs, to be executed at a higher priority.
The workload management enhancements include
• Workload queues
• Configurations-switchable sets of workload queues (for example, a configuration for month-end processing)
• Session level workload management
• Real time monitoring of workloads
• Auditing of workloads
To set up the workload management system, the DBA can enter values into three new tables in the iidbdb database or use the ALTER WLM statement to create the workload management configurations.
New SQL elements include:
• ALTER WLM statement to change the configuration of workload management. ALTER WLM can be used while the system is running.
• New request names on the DBMSINFO function, WLM_ENABLED, WLM_CONFIG, and WLM_AUDIT, to view information on the workload management system.
For more information, see
Managing Workload on page 217 and the
SQL Language Guide.
Scalar User-defined Functions Enhancements
Enhancements to JavaScript and Python scalar user-defined functions allow the code to be run in an environment that is secure, where access to critical resources is tightly controlled.
The enhancements let you:
• Run Python UDFs securely.
• Run JavaScript and Python UDFs in a safe, sand-boxed environment in a container.
• Use UDFs in production by sand-boxing.
• Easily install the required version of Python, which is included in containerized UDF.
• Use popular external Python libraries. The pre-built container bundles Python and JavaScript libraries and datasets.
Changes for this feature include a new X100 configuration file, udf_engine.conf, that defines available UDF engines and their properties.
For more information, see the SQL Language Guide and System Administrator Guide.
Encryption Key Management
The key management feature lets you manage encryption keys for data at rest encryption. It includes:
• ALTER PASSPHRASE statement-Changes the passphrase, which automatically changes the database key but leaves the table keys untouched.
• ALTER KEYS statement-Changes the table keys without changing the passphrase and database key.
For more information, see the Security Guide and SQL Language Guide.