Systems Management Associates

Cloud Database

Foglight for Azure Cosmos DB

Request units, latency, and availability

Description

Azure Cosmos DB is Microsoft's globally distributed, multi-model NoSQL database service. Teams choose it for elastic throughput measured in request units (RUs), single-digit millisecond reads and writes at any scale, and turnkey multi-region replication for mission-critical applications.

Foglight for Azure Cosmos DB is an SMA cloud-API cartridge. Its FglAM agent authenticates with an Azure AD service principal and collects Azure Monitor metrics together with control-plane inventory from the Cosmos DB resource provider - databases, containers, regions, and provisioned throughput - without Cosmos account keys or JDBC.

Business Challenge

Cosmos DB exposes rich metrics in Azure Monitor, but throttling (HTTP 429), RU saturation, rising server-side latency, and storage growth are easy to miss when each signal lives in a separate portal blade. Platform teams need history, roll-up health, and alerting tuned to how Cosmos actually behaves under load.

Performance, storage, and availability monitoring are critical to enterprise database management. Maintaining tooling to perform automated monitoring can be both challenging and distracting for teams whose focus is on developing and operating database applications - especially when Cosmos accounts span multiple regions and shared-throughput databases.

Key Features

Foglight for Azure Cosmos DB is account-keyed: the CosmosAccount is the top-level topology object, with databases, containers, and regions rolling up beneath it. Twelve purpose-built dashboards span Overview, Throughput, Requests & Latency, Storage, Databases, Containers, Regions, Workload, Resource, I/O, Performance, and Contention.

The agent uses OAuth2 client credentials against Azure AD, then polls Azure Monitor for metrics and the DocumentDB ARM resource provider for inventory. Service-principal tenant, client ID, secret, subscription, resource group, and account name are agent properties. A comprehensive ruleset covers availability, RU saturation, throttling, latency, storage, gateway CPU, cache hit rate, and server errors.

As with all Foglight solutions, Foglight for Azure Cosmos DB runs as a native Foglight process on the Foglight Agent Manager (FglAM) and leverages the Foglight Rules Engine to alert on potential issues before they significantly affect performance and availability.

Ready to see Azure Cosmos DB in action?

Request a free trial or talk to our engineers about your environment.