Skip to main content

Overview

The Ops Platform monitors and diagnoses HAP/HDP private deployment environments. It brings together host, middleware, and Kubernetes cluster monitoring information, with log search, tracing, slow query analysis, and alerts to help operations teams check status and troubleshoot problems.

After deployment, add hosts and middleware in Data Sources, then connect service logs and tracing as needed for your environment.

Feature Overview​

Ops Platform overview

ModuleCapabilityDocumentation
OverviewSystem health, monitoring task statistics, active alerts, and alert activity over the last 24 hoursThis page
Resource MonitoringView monitoring information for hosts, middleware, and Kubernetes clusters; embed external pages such as Flink as custom pagesResource Monitoring
AlertsConfigure monitoring rules and notification channels; view alert status, recovery, and historyAlert Rules
LogsSearch container logs and HAP/HDP microservice logs by time and keywordLog Search
TracingView request paths and duration across microservices in Kubernetes environmentsTracing
Slow Query DiagnosticsAnalyze MongoDB slow queries, inspect query details, and review index recommendationsMongoDB Slow Query Analysis

When deployed inside the monitored Kubernetes cluster, the platform can collect that cluster's metrics automatically. To monitor other clusters, deploy collection components in each target cluster. See Kubernetes Cluster Monitoring.

Technical Architecture​

The Ops Platform uses Prometheus, Grafana, Loki, Tempo, and Grafana Alloy to collect, store, and display monitoring data:

Ops Platform architecture

  • Metrics: Prometheus collects and stores host, middleware, and Kubernetes cluster metrics. Grafana provides monitoring dashboards.
  • Logs: Alloy collects container logs, while HAP/HDP microservices push service logs to Loki. Configure service log ingestion on the product services. See Service Logs.
  • Traces: Applications report traces to Alloy through OTLP. Alloy forwards them to Tempo for storage, and Grafana displays them.
  • Alerts: The platform queries metrics or checks monitored targets according to rules and sends alerts through notification channels.
  • Slow queries: The platform collects MongoDB slow query records, analyzes them, and provides index recommendations.

Alert rules, status, history, and slow query history are stored in the platform's own MongoDB, separately from the monitored business databases.

Component Versions​

The platform uses the following components and versions:

ComponentVersionPurpose
Prometheus3.1.0Metrics collection and storage
Grafana12.2.1Unified visualization UI
Loki3.5.8Log aggregation and search
Tempo2.9.0Distributed trace storage
Grafana Alloy1.11.3Unified observability collection (container logs/OTLP)
Node Exporter1.8.2Host resource metrics
kube-state-metrics2.10.1Kubernetes cluster object status
Kafka Exporter1.6.0Kafka metrics collection
Elasticsearch Exporter1.5.0Elasticsearch metrics collection
Redis Exporter1.44.0Redis metrics collection
MySQL Server Exporter0.15.1MySQL metrics collection
MongoDB Exporter0.42.0MongoDB metrics collection
MongoDB4.4.28Alert subsystem storage, independent of monitored targets