Managed OpenTelemetry backend

Run one OpenTelemetry agent. Query it in Insights.

Enroll the OpenTelemetry agent and the portal manages its config over OpAMP. Metrics, logs and traces land in Insights, where the explorer, dashboards, snapshots and alerting already live. The Prometheus-, Loki- and Tempo-compatible APIs stay open, so a Grafana you already run keeps working.

enroll an agentopamp extensionX-Scope-OrgID401 mismatchimport a dashboardcardinality limits

Start here

The xScaler OpenTelemetry agentRecommended

A standard OpenTelemetry Collector under OpAMP management. You write one block of YAML to enroll it. Every pipeline after that comes from the portal, selected by the agent's labels.

1

Create an enrollment token

Fleet Management → Enrollment. Stamp default labels such as environment and region onto every agent the token enrolls. The token is shown once.

2

Point the agent at OpAMP

The agent registers, then trades the enrollment token for its own credential. Reconnects use that credential.

3

Assign a config template

Label selectors decide which agent runs which pipeline. The portal shows the effective config and the delivery history per agent.

otel-collector-config.yaml

# the only file you edit by hand
extensions:
  opamp:
    server:
      ws:
        endpoint: wss://agents.xscalerlabs.com/v1/opamp
        headers:
          Authorization: "Bearer <enrollment-token>"

service:
  extensions: [opamp]

Prometheus remote_write, Grafana Alloy and the OTel SDKs write to the same endpoints with the same two headers. eBPF instrumentation gets traces without touching your code.

Send data

Three signals, one set of credentials

One token and one tenant header cover all three. The agent sends every signal from a single config. Each signal keeps its own write endpoint and its own query language.

MetricsPromQL

Prometheus-compatible ingestion and query.

OpenTelemetry agentPrometheus remote_writeGrafana Alloy

LogsLogQL

Structured and plain-text lines, Loki-compatible.

OpenTelemetry agentGrafana AlloyOTel SDKs

TracesTraceQL

Distributed spans across services, Tempo-compatible.

OpenTelemetry agentOTel SDKseBPF auto-instrumentation

Read your data

InsightsIn the portal

An explorer, dashboards, snapshots and alerting come with the tenant. There is nothing to host and no data source to wire up before the first query.

Metrics

Break a metric down by label and read the result, with no PromQL to write first.

Logs

Filter a stream by label, then open the trace a line belongs to.

Traces

Span tree, critical path and service graph for one request.

Explorer

Several queries at once, each with its own signal, so logs sit beside the metric that raised the question.

Dashboards

Visual panel editor with variables, annotations and exemplars. Grafana JSON imports as-is.

Snapshots

Freeze what is on screen and send the link to whoever is on the call.

Alerting

Alert rules, contact points, notification policies, silences and mute timings.

Your own Grafana

The Prometheus-, Loki- and Tempo-compatible APIs stay open, so existing dashboards keep reading.

Config from the portal

Edit a pipeline once. OpAMP delivers it to every agent whose labels match the assignment.

Insights is included

Explorer, dashboards and alerting come with the tenant. There is no Grafana to run to read your own data.

Regional by default

Tenants are provisioned in a region and every hostname follows it.

MCP on every plan

Point Claude Code or Cursor at your telemetry. No key to create.

By task

What do you want to do?

Getting data in

Enroll the OpenTelemetry agentOpAMPSend config to a fleet of agentsTemplatesI already run my own OTel CollectorOTLPI already run Prometheusremote_writeI already run Grafana AlloyAlloyI want to instrument my service codeOTel SDKsI want traces without touching my codeeBPF

Looking at your data

Break a metric down by labelInsightsBuild a dashboardInsightsImport a Grafana dashboardInsightsSet up an alert ruleAlertingShare what I am looking atSnapshotsPoint my own Grafana at xScalerData sourcesRun a query from the command lineQuery API

Running the platform

Create a tenantPortalMint and rotate an API tokenPortalCheck what a tenant is ingestingUsageAutomate all of itAPI reference

When something is wrong

I get 401 x-scope-orgid mismatchAuthMy writes are being rejectedLimitsMy data is not showing upTroubleshootingI need to reduce cardinalityLimits

AI

Two ways AI shows up here

Point a coding agent at your telemetry, or measure the AI you already ship.

xScaler & AI tools

An MCP server your coding agent queries directly for metrics, logs and traces. Every answer reports how much data it scanned.

Connect a client →

AI observability

Your LLM calls as OpenTelemetry GenAI spans and metrics: request rate, latency, and token usage per model and operation.

Instrument your app →

Monitor your stack

Ready-made integrations

Each integration ships a collector config and a starter dashboard. Search the catalogue by name.

KubernetesPostgreSQLRedisNGINXKafkaMongoDBClickHouseLinuxAWS CloudWatchIstioVaultBrowse all 100+ →