New and changed in Vertica 12.0.3

• 1: Client connectivity
• 2: Containers and Kubernetes
• 3: Machine learning
• 4: Partitioning
• 5: Query optimization
• 6: Security and authentication
• 7: SQL functions and statements

1 - Client connectivity

Client label environment variable

You can now set a label for the vsql client with the VSQL_CLIENT_LABEL environment variable. For details, see vsql environment variables.

2 - Containers and Kubernetes

TLS certificate authentication to prometheus metrics

In some environments, you might need to access Prometheus metrics from an external client. Vertica provides the prometheus.tlsSecret Helm chart parameter to configure a role-based access control (RBAC) proxy sidecar to authenticate requests with user-provided TLS certificates.

For details, see Helm chart parameters and Prometheus integration.

Set readiness probe on container

The readinessProbeOverride custom resource definition (CRD) parameter overrides settings for the default readiness probe so that you can fine-tune when the Vertica pod is ready to accept traffic.

For details, see Custom resource definition parameters.

IRSA profile authentication to Amazon EKS

You can authenticate to Amazon Elastic Kubernetes Service (EKS) with IAM roles for service accounts (IRSA). For details, see Configuring communal storage

Operator scheduling rules for helm chart

You can control which node the operator pod is scheduled on with the following Helm chart parameters:

• affinity

• nodeSelector

• priorityClassName

• tolerations

For details, see Helm chart parameters.

Set liveness and startup probes on container

The livenessProbeOverride and startupProbeOverride custom resource definition (CRD) parameters override settings for the corresponding default probes. These parameters fine-tune how the container and the Vertica process within the container indicate their state to other objects in the StatefulSet.

For details, see Custom resource definition parameters.

Override pod-level security context

The podSecurityContext custom resource definition (CRD) parameter can elevate pod-level privileges so that you can perform privileged actions, such as setting sysctl commands in the pod.

For details, see Custom resource definition parameters. For additional details about pod-level privileges, see the Kubernetes documentation.

3 - Machine learning

Extract predictor importance of XGBoost models

The new XGB_PREDICTOR_IMPORTANCE function measures the importance of the predictors in XGB_CLASSIFIER and XGB_REGRESSOR models.

K-prototypes

You can now use the k-prototypes clustering algorithm to cluster mixed data into different groups based on similarities between data points.

New MLSUPERVISOR role

A new predefined role MLSUPERVISOR can delegate DBADMIN privileges for ML model management to other users.

4 - Partitioning

New drop partitions (d) lock

DROP_PARTITIONS now sets an exclusive D lock on the target table. This lock is only compatible with I-lock operations, so table load operations such as INSERT and COPY are allowed during the drop operation. All other operations that require a table lock, such as UPDATE or DELETE, are blocked until the drop partition operation is complete and the D-lock is released.

5 - Query optimization

Disabling projections

ALTER PROJECTION now supports a new DISABLE parameter, which marks a projection as unavailable to the optimizer.

6 - Security and authentication

Authentication filtering

Users can now specify the authentication record they want to authenticate with by providing the credentials required by that record. Previously, users could only authenticate with the highest priority authentication record or the records to which it fell through. For details, see Authentication filtering.

7 - SQL functions and statements

LIKE support for ANY/ALL

LIKE operators now support the keywords ANY and ALL, which let you specify multiple patterns to test against a string expression. For example, the following query qualifies the case-insensitive ILIKE with ALL to find all customer names that contain the strings media and ever:

=> SELECT DISTINCT (customer_name) FROM customer_dimension
     WHERE customer_name ILIKE ALL ('%media%','%ever%') ORDER BY customer_name;
 customer_name
---------------
 Evermedia
(1 row)

For details, see LIKE ANY/ALL Usage.