New and changed in Vertica 23.4

• 1: Data export and replication
• 2: Data load
• 3: Database management
• 4: Machine Learning
• 5: Management Console
• 6: Security and authentication
• 7: Stored procedures
• 8: Upgrade and install

1 - Data export and replication

Initiate server-based replication from the source database

You can now initiate server-based replication from the source database. The replication steps are similar to target-initiated replication, except that you must instead connect to the target database from the source database and specify the target database in a TO clause when calling REPLICATE. For more information, see Server-based replication.

2 - Data load

Automatically load new files

A data loader automatically loads new files from a location, so that you do not have to add them to Vertica manually. Automatically loading new data into ROS tables is an alternative to using external tables and can save on API costs for object stores.

A data loader is tied to a path for data and a target table. When executed, the loader attempts to load files that it has not previously loaded. A loader has a retry limit to prevent malformed files from being tried over and over. Each loader records monitoring information in an associated table.

To run a data loader periodically, you can use a scheduled stored procedure to execute the loader.

For details and an example, see Automatic load.

ORC parser supports loose schema matching

By default, the ORC parser uses strong schema matching. This means that the load must consume all columns in the data and in the order they occur in the data. You can, instead, use loose schema matching, which allows you to select the columns you want and ignore the rest. Loose schema matching depends on the names of the columns in the data rather than their order, so the column names in your table must match those in the data. Types must match or be coercible. Loose schema matching for ORC behaves the same way as it does for Parquet. For details on how to use loose schema matching, see the ORC reference page.

Partitioned paths

Vertica previously supported partition pruning for Hive-style partitioned data. Vertica now supports loading and pruning from any partitioned path. For example, given paths like /data/2023/01, you can now read the year and month values from the path, and at query time Vertica automatically skips reading partition directories that are not needed. See Partitioned data.

3 - Database management

Endpoint authentication

The following documentation endpoints no longer require authentication:

• NMA endpoints:
  • /api-docs/
  • /v1/health
• HTTPS service endpoints:
  • /v1/version
  • /swagger/ui
  • /swagger/{RESOURCE}
  • /api-docs/oas-3.0.0.json

4 - Machine Learning

Support for chi-square independence test

With the new CHI_SQUARED function, you can compute the conditional chi-square independence test on two categorical variables to find the likelihood that the two variables are independent. The function also supports the option to condition the test on another set of variables.

ARIMA models support differencing

ARIMA models in Vertica now support differencing. This operation can remove or reduce trends in time series data. To train an ARIMA model that applies differencing, set the integration parameter d to an integer between one and ten. This value specifies the difference order of the model, which determines how many times the differencing operation is applied to the input data.

For an example that trains an ARIMA model that uses differencing, see ARIMA model example.

5 - Management Console

Custom image on Google Cloud Platform

You can create a custom Management Console (MC) image for Google Cloud Platform (GCP). Create an instance with the published MC image, and then add dependencies or change environment settings on that instance. After you make changes, you can create a new image that includes the published MC instance and the new dependencies or settings. You can use the new image when you create or revive a subcluster.

For details, see Custom GCP image.

Home page redesign

The Management Console (MC) home page has a new design that provides a central location to view and act upon critical information about the databases that the MC manages.

After you log into the MC, you go to the Databases page, which has an interactive dashboard that displays important details for each database and subcluster. You have options to perform database actions, and the dashboard graphics link to the relevant section of the MC.

For details, see Management Console home page.

Toolbar and navigation redesign

The toolbar and navigation use a new design to provide an enhanced and consistent user experience.

• The toolbar provides quick access to alerts, the language selector, Vertica resources, and user actions.
• MC navigation is now a multi-level pane that provides access to system-level and section-specific navigation options. You can access the system-level options from anywhere within the MC, and the section-specific options are displayed for each system-level option.

For details, see Management Console toolbar and navigation.

6 - Security and authentication

CONNECT TO VERTICA: Passwordless authentication

You can now omit the user's password in the call to CONNECT TO VERTICA, authenticating to the target database with one of the following methods instead:

• Credential forwarding: Forward the password for the current user to the target database
• TLS authentication: Authenticate with TLS

For details, see CONNECT TO VERTICA

7 - Stored procedures

OUT and INOUT parameter modes

You can now return values as a result set from stored procedures with OUT and INOUT parameters for non-complex data types. For details, see Parameter modes.

8 - Upgrade and install

Skip RPM copy when running the installation script

The installation script now supports a no-rpm-copy option that bypasses the time-consuming step of copying the RPM to each node in the cluster. The RPM must be present on each node specified by --hosts, and you must provide the path to the local RPM files with the --rpm-path option. For details, see --no-rpm-copy.