Single service comparison

Azure Monitor consolidates metrics, logs, traces, and alerts under one data plane with a unified workspace model and KQL; Huawei splits these across CES (platform metrics), LTS (log ingestion and query), APM (distributed tracing and RUM), and AOM (unified dashboards, container insights, and hosted Prometheus). AOM 2.0 aggregates data from CES, LTS, APM, and Prometheus into shared dashboards, but each underlying service retains its own console, API, billing, and query syntax, so there is no single workspace or query language equivalent to KQL.

API and data model differ materially. Azure exposes a REST ingestion API, Azure Resource Graph, and KQL across one workspace; CES uses per-service metric APIs and second-precision metrics, LTS uses its own log search syntax, and APM uses proprietary Java agents and topology APIs. Retention, RBAC, and cross-resource queries are workspace-level in Azure but service- and region-level on Huawei, so cross-workspace correlation and identity isolation require explicit per-service configuration rather than a single workspace boundary.

Scaling, HA, and integration boundaries differ. Azure Monitor is regionally deployed with global Log Analytics and Azure Advisor integration; CES, LTS, APM, and AOM are independently regional with separate quotas and collectors. AOM interconnects with CCE and hosted Prometheus via remote write and PromQL, while CES triggers Auto Scaling and LTS works with CTS for audit logs. Operational responsibility is therefore distributed across four Huawei services rather than one control plane, increasing configuration surface for parity validation.

Assessment and target choice: inventory Azure Monitor telemetry by signal type. Platform resource metrics and health map to CES; application and custom logs to LTS; distributed traces, RUM, and transaction insights to APM; and unified dashboards, container/CCE observability, and Prometheus workloads to AOM. Confirm per-service regional availability and quotas for each target region, since there is no one-to-one workspace equivalent and feature gaps vary by region.

Data and config migration: there is no Huawei-provided one-click importer; replan ingestion. Recreate metric dashboards and alarm rules in CES/AOM using native templates, translate KQL log queries to LTS search syntax and rebuild retention policies per LTS tier, and re-instrument applications with APM agents or OpenTelemetry exporters (AOM/Prometheus remote write). Re-export Azure activity logs to LTS for audit continuity via CTS-style ingestion. No automated schema conversion exists, so expect query and alert re-authoring.

Validation and cutover: run dual-parallel telemetry on Azure and Huawei during a shadow period, comparing metric accuracy, alert latency, and trace completeness. Validate alarm action mappings (Huawei SMN topics replace Azure action groups) and dashboard parity before switching consumers. Confirm RBAC via Huawei IAM roles per service, since workspace-level RBAC does not transfer. Cutover per workload only after parity checks pass.

Cost model and gaps: Azure bills ingested/retained telemetry, query volume, and alert execution; Huawei bills CES, LTS, APM, and AOM ingestion, retention, analysis, and alerting separately, plus MQTT-free SMN notifications. Recompute TCO with peak ingest, retention tiers, query volume, and cross-region traffic. Note gaps: no single cross-service query language, weaker cross-workspace correlation, and limited managed equivalents for some Azure features such as workload monitoring and Azure Advisor recommendations require custom alternatives.