Overview of Datadog architecture, components, data collection and delivery models, observability features, integrations, agents, data flow, and enterprise networking and site considerations for deployment
In this lesson we examine Datadog’s architecture and the platform components that make the observability solution work end-to-end.Although Datadog is presented as a single Software-as-a-Service (SaaS) console, the platform is composed of several integrated parts: local collectors (Agents), cloud and system Integrations, and APIs that enable custom logic, ingestion, and checks. These components work together to collect, process, store, and surface telemetry (metrics, logs, traces, and profiles).
When evaluating observability platforms you’ll encounter two common delivery models:
Delivery model
Typical use case
Notes
SaaS
Fast onboarding, managed backend
Datadog is primarily a SaaS offering — the console and backend are hosted by Datadog. Agents and integrations run in your environment to collect data.
Self-hosted (on-prem)
Strict regulatory/compliance or network constraints
You host the control plane and storage (e.g., self-hosted Grafana). Greater operational overhead but more control.
Choose based on technical, compliance, and operational requirements: latency, data residency, control of infrastructure, and security posture are common decision drivers.
Datadog Console (SaaS): Central UI, dashboards, alerting, and control plane.
Agents: Lightweight collectors that gather metrics, logs, traces, and continuous profiles from hosts, containers, and serverless runtimes.
Integrations & APIs: Cloud, database, and third-party system connectors; public APIs for custom ingestion, checks, and automation.
Clients: Engineers and responders use browsers and mobile apps to view dashboards, receive alerts, and manage incidents.
Table — Core components at a glance:
Component
Role
Examples
Console (SaaS)
UI and control plane
Dashboards, monitors, Incident Management
Agent
Local telemetry collection
Host agent, Containerized Agent, APM tracer
Integrations
Connector to services and platforms
AWS, Azure, Kubernetes, Databases
APIs
Programmatic ingestion and automation
Custom metrics API, Events API, Checks
Datadog supports telemetry from containerized applications, serverless functions (AWS Lambda, Azure Functions), cloud platforms, on‑prem infrastructure, and frontend applications. Each source typically has a tailored collection method or a dedicated integration.
Data collection typically begins with Agents and platform integrations:
Agents: Installed where telemetry originates (hosts, containers, or as binaries). For Kubernetes, deploy an Agent inside the cluster to collect node, pod, and service telemetry.
Serverless and cloud-native integrations: Some integrations collect telemetry without a persistent agent by using platform APIs or function-level instrumentation (e.g., Datadog’s Lambda Forwarder for AWS Lambda logs).
Ingestion pipelines: Datadog supports pre-ingestion processing, enrichment, and parsing rules so telemetry is normalized before storage and analysis.
Example: installing the Datadog Agent into a Kubernetes cluster (Helm):
Datadog is accessed over the public internet with standard browsers and the mobile app.
Integrate with your Identity Provider (IdP) for SSO and centralized access control to enforce corporate policies.
Egress, proxies, and firewall considerations:
Many organizations require outbound traffic to be routed through proxies or firewalls. Configure Datadog Agents and integrations to work through your proxy.
Verify required allowlists (hostnames, IPs) and TLS interception rules so agents can reach Datadog endpoints.
Datadog operates multiple sites/regions. Your site selection affects latency and compliance (data residency). Important points:
Site selection is persistent for an organization; data cannot be moved later.
Evaluate regulatory and business requirements (e.g., GDPR, data residency) before choosing a site.
Carefully evaluate data residency and compliance requirements before choosing your Datadog site. The selection is persistent for your organization and cannot be changed later.
Datadog is a SaaS console backed by a distributed collection layer (Agents and Integrations) and public APIs. Understanding where and how telemetry is collected, the difference between agent-based and platform-native integrations, and the implications of networking and site selection will help you design an observability deployment that meets performance, security, and compliance goals.Further reading and references:
