Containerized Metadata Persistence
The Containerized Metadata Persistence (CMDP) feature preserves metadata so that recovery from container or pod failures is possible by restarting the cluster, without the need to reinstall analytic functions on restart.
Failure Detection Recovery
Failure Detection Recovery (FDR) detects failures and attempts to recover from those failures automatically, thereby reducing cluster downtime.
Function Execution
ML Engine function execution feature enables Advanced SQL Engine to run functions that are remotely available on ML Engine.
In Memory Analytics Tables
The In Memory Analytics Tables (IMAT) enhance performance by keeping data in memory and reducing data copies.
Kubernetes
Kubernetes is a container orchestration platform. The applications that comprise ML Engine run in containers. Kubernetes launches and operates the container infrastructure that makes up the ML Engine.
Query Monitoring
Query monitoring helps administrators measure compute resource utilization. Query monitoring is useful in identifying expensive phases of a query that tax certain resources.
Event Management Events
Event Management (EM) events for ML Engine help administrators detect and manage situations where there is cluster degradation and down incidents.
System Monitoring
Teradata Vantage can monitor ML Engine worker node (pod) resource consumption, correct any resource bottlenecks, and pinpoint long-term and cyclical utilization trends, as well as set health thresholds.
Workload Management
Workload Management (WLM) manages workload performance by prioritizing workload requests, monitoring system activity, and acting when predefined limits are reached.