How to Migrate from Azure Anomaly Detector to Canary Edge (October 2026 Deadline)

When Does Azure Anomaly Detector Shut Down?

Azure Anomaly Detector has a hard retirement date of October 1, 2026. Microsoft has confirmed there is no extension period and no legacy mode. On that date, the service goes dark permanently.

What stops working on October 1, 2026:

• All API endpoints under *.cognitiveservices.azure.com/anomalydetector/* stop responding
• All trained multivariate models stored in the Azure service are deleted
• All stored configurations and resource settings are destroyed
• The Azure portal UI for Anomaly Detector is removed

This is not a soft deprecation. Teams that have not migrated by October 1, 2026 will experience immediate production outages. Any pipeline, alerting system, or application that calls the Azure Anomaly Detector API will fail with connection errors once the service is retired.

Microsoft is not offering data export options for trained models. Because Canary Edge trains models automatically from your time-series data, no model migration is required — only a configuration change.

What Code Changes Are Required to Migrate?

The answer is two lines. That is the entire code change required for the vast majority of teams migrating from Azure Anomaly Detector to Canary Edge.

Line 1: Update the endpoint URL

Before: https://[your-resource].cognitiveservices.azure.com After: https://api.canaryedge.com

The path structure remains identical. If your code calls /anomalydetector/v1.1/timeseries/entire/detect, it continues to call that same path on the Canary Edge endpoint.

Line 2: Update the API key

Before: your Azure Cognitive Services subscription key After: your Canary Edge API key (starts with cnry_live_)

The header name stays the same: Ocp-Apim-Subscription-Key. No header rename needed.

That table represents the complete migration surface. Everything else — your request payloads, response parsing logic, downstream alerting, data pipelines — stays exactly as it is.

Is the Canary Edge API 100% Compatible with Azure?

Yes. Canary Edge is designed as a drop-in replacement for Azure Anomaly Detector. The compatibility is complete at the request level, the response level, and the SDK level.

Request format: Identical. Canary Edge accepts the same JSON body: a series array of timestamp/value pairs, plus granularity and optional parameters like sensitivity, maxAnomalyRatio, and period. No request schema changes are required.

Response format: Identical. Canary Edge returns the same response fields: - isAnomaly — boolean array, one per data point - expectedValues — model-predicted values for each point - upperMargins and lowerMargins — confidence interval margins - severity — anomaly severity score (0.0 to 1.0) - period — detected period of the series - isPositiveAnomaly and isNegativeAnomaly — directional flags

Your response parsing code, downstream logic, and alert thresholds all continue to work without modification.

SDK compatibility: The official Azure Anomaly Detector SDK for Python, JavaScript, C#, and Java all work with Canary Edge. These SDKs accept a configurable endpoint URL at initialization time. Pointing the SDK at https://api.canaryedge.com and supplying a Canary Edge API key is the only change required. Every SDK method — detect_entire_series, detect_last_point, detect_change_point, and all multivariate methods — works identically.

Step-by-Step Migration from Azure Anomaly Detector

Follow these six steps to complete a production migration from Azure Anomaly Detector to Canary Edge:

Step 1: Sign up at canaryedge.com Create a Canary Edge account at canaryedge.com/sign-up. No credit card is required for the free tier.

Step 2: Get your API key From the Canary Edge dashboard, navigate to API Keys and generate a new key. Your key starts with cnry_live_. Copy it securely — treat it like any production secret and store it in your secrets manager or environment variable system.

Step 3: Update the endpoint URL In your application configuration, replace the Azure endpoint https://[your-resource].cognitiveservices.azure.com with https://api.canaryedge.com. If you use environment variables (recommended), this is a one-line change in your config file or secrets manager.

Step 4: Update the API key Replace your Azure Cognitive Services subscription key with your Canary Edge API key. The header name Ocp-Apim-Subscription-Key does not change.

Step 5: Test with one machine or data stream Run your existing test suite against the Canary Edge endpoint. Send a representative time series and verify the response structure matches your expectations. Compare isAnomaly arrays to confirm the detection output is reasonable for your data.

Step 6: Roll out to all machines Once testing passes, update all production instances. Because no downstream code changes are required, rolling out to additional machines or data streams is a configuration push, not a code deployment.

Total calendar time for most teams: one day. If your infrastructure uses Kubernetes ConfigMaps, Terraform variables, or a secrets manager like AWS Secrets Manager or HashiCorp Vault, the actual change is a single key-value update per environment.

How Does Canary Edge Pricing Compare to Azure Anomaly Detector?

Azure Anomaly Detector used consumption-based pricing starting at approximately $0.314 per 1,000 API transactions for the S0 tier. At scale, this meant unpredictable monthly bills that varied with data volume. Teams monitoring many machines or running high-frequency detection pipelines faced significant cost variance.

Canary Edge uses flat-rate pricing:

Free tier: 10,000 data points per month. Suitable for evaluation, development, and small deployments. No credit card required.

Pro tier ($49/month): 1,000,000 data points per month. Covers most production deployments — a 50-machine facility sampling every 10 seconds generates approximately 432,000 data points per day, fitting comfortably within the Pro tier.

Enterprise: Custom pricing for high-volume deployments, on-premise requirements, SLA guarantees, and dedicated infrastructure. Contact sales at canaryedge.com/contact.

The shift from consumption pricing to flat-rate pricing simplifies budgeting. Your monthly cost is predictable regardless of data volume spikes, backfill operations, or increased monitoring frequency. For teams that ran Azure Anomaly Detector at meaningful scale, the Pro tier at $49/month typically represents a significant reduction in monthly spend compared to Azure's consumption rates.

What Is the Migration Timeline?

With an October 1, 2026 hard deadline, teams should work backward from that date to allow adequate testing and buffer time.

April 2026 (now): Start Sign up for Canary Edge, generate API keys, and run initial compatibility testing. Identify all services and pipelines that call Azure Anomaly Detector — this inventory is the critical first step. Most teams discover they have more integrations than they initially realize.

May–June 2026: Development and staging migration Update development and staging environments to use Canary Edge. Run your full test suite. Validate detection quality on representative data from your production machines. This phase typically surfaces any edge cases in response handling.

July 2026: Production migration Migrate production workloads to Canary Edge. Running parallel validation — calling both Azure and Canary Edge with the same data and comparing outputs — gives teams confidence before fully cutting over.

August 2026: Full cutover Complete the migration and decommission Azure Anomaly Detector resources. Removing the Azure dependency before the deadline gives you time to address any unexpected issues.

September 2026: Buffer month Keep this month free of planned migration work. Use it as a safety buffer for any teams, pipelines, or integrations that took longer than expected. September is not a migration month — it is a contingency month.

October 1, 2026: Azure Anomaly Detector retires Teams who completed migration by August are fully protected. Teams still on Azure on this date will experience immediate production outages.

Starting in April gives you five months of working time. Most teams complete the technical migration in one to five days. The remaining time is for testing, validation, and organizational processes like change management approvals and vendor notifications.