Artificial intelligence news.
Straight from the original announcements.

Dynatrace adds Google Cloud observability with Pub/Sub risk detection

November 19, 2026, 17:31By the AI News Bites Editorial TeamSource: Dynatrace
Dynatrace adds Google Cloud observability with Pub/Sub risk detection
Image: Dynatrace

Dynatrace says Google Cloud observability is now generally available in Dynatrace Clouds, bringing resource configuration, metrics and logs into a shared investigation context. The company is using a Pub/Sub incident example to show how teams can compare message age with retention settings and identify risks across connected Google Cloud projects.

Key points
  • Google Cloud observability is now generally available in Dynatrace Clouds.
  • Dynatrace combines Google Cloud configuration, metrics and logs in one context.
  • A DQL query compares Pub/Sub message age with configured retention.
  • The example found 1,932 refunds in a dead-letter queue.
  • The dead-letter subscription retained messages for seven days.

A single view of Google Cloud resources

The release extends Dynatrace's Cloud Observability experience to Google Cloud. According to Dynatrace, users can analyze configuration, metrics and logs with Dynatrace Query Language, or DQL, while existing cloud labels can route alerts to resource owners.

The company's example focuses on an online fashion retailer whose settlement job publishes approved refunds to a Pub/Sub topic. A push subscription delivers the messages to a Cloud Run service, which pays refunds through a payment provider.

Why the dead-letter queue hid the retention risk

In the representative incident, a larger-than-usual batch caused the refund service to reject work for half an hour. The Cloud Run service reached its ceiling of 20 instances and returned 429 responses. Dynatrace opened a problem at 10:26, and the limit was increased from 20 to 60 at 10:43. The last 429 occurred at 10:54.

The subscription had a dead-letter policy that moved a message after five failed delivery attempts to a separate topic and subscription. Pub/Sub retained the messages there for seven days, but the dead-letter subscription had no consumer and no alert. Dynatrace says 1,932 refunds were forwarded between 10:30 and 10:55.

That setup prevented immediate data loss, but it shifted the risk to retention. As the source explains, “After forwarding an undeliverable message, the Pub/Sub service removes the message from the subscription.” Monitoring the original subscription therefore no longer showed the forwarded messages, while a simple depth alert would only show that something was waiting, not when it could expire.

Combining message age with configuration

Dynatrace's proposed approach compares the oldest unacknowledged message age with the retention configured for the relevant subscription. Retention is configuration rather than a metric, but the company says DQL can combine the two values because subscription configuration is available alongside monitoring data in Smartscape and stored together in Grail.

In the example, the query calculates a ratio by dividing the oldest message age by the subscription retention. On refunds-push, the ratio peaked at 0.0016, equivalent to about sixteen minutes against seven days. The backlog did not age because failed messages were retried and then forwarded to the dead-letter subscription.

What this means for cloud operations teams

The example shows why a dead-letter queue can make an incident appear resolved while leaving a separate deadline to track. Dynatrace's workflow is designed to follow relationships between resources, compare operational data with configuration, and turn the result into an alert covering connected projects.

For teams managing Pub/Sub subscriptions, the relevant signal is therefore the time remaining before retention, rather than queue depth alone. In the scenario, the incident review correctly found no data loss, but replay was postponed to the next sprint, leaving the seven-day retention window as the key condition to monitor.

DynatraceGoogle CloudGoogle Cloud Pub/SubDynatrace Query Languagecloudenterprise AI

Related news

Dynatrace earns AWS Data Competency for AI-ready cloud operations
BusinessNovember 19, 2026, 22:18

Dynatrace earns AWS Data Competency for AI-ready cloud operations

Dynatrace has achieved the AWS Data Competency, a recognition the company says validates its expertise in building unified, AI-ready data foundations on AWS. The…

Google Cloud adds object enrichments for agent-ready data
Products and toolsNovember 19, 2026, 18:00

Google Cloud adds object enrichments for agent-ready data

Google Cloud has introduced object enrichments, a Cloud Storage capability that uses AI models to extract contextual metadata from unstructured files and store it…

Dell expands enterprise AI ecosystem with OpenAI and 10 other partners
BusinessNovember 16, 2026, 15:00

Dell expands enterprise AI ecosystem with OpenAI and 10 other partners

Dell Technologies is expanding its AI Ecosystem Program with joint enterprise solutions from OpenAI and other technology providers across coding, security, data…