Release Announcement: Vald v1.7
2 min readJan 25


A new version of v1.7, a minor update, has been available since this week.

This update introduces several new features:

And has breaking changes:

  • Jaeger, OpenCensus, and StackDriver for tracing and metrics are out of the Vald cluster

In this post, we will explain how to use new metrics & tracing with OpenTelemetry.

Observing with OpenTelemetry Protocol

Since v1.6, Vald has supported OpenTelemerty. From v1.7, Vald fully migrated to OpenTelemetry for metrics, logs, and traces. According to it, all metrics and tracing configuration dependencies except OpenTelemetry are removed from Vald.

Because of it, the Helm chart has breaking changes, so please be careful when upgrading your current Vald cluster.

Here is the new architecture of metrics & tracing.

New Architecture for Observing with OpenTelemetry Protocol

The data flow is as follows:

  • All Vald components send their metrics data to OpenTelemetry Collector using OpenTelemetry Protocol.
  • OpenTelemetry Collector pushes collected data to any monitoring platform, e.g., Grafana, Jaeger, etc.

It enables you to use any observability tools which comply the OpenTelemerty.

FYI: shows the native support vendors.

Observing the Vald cluster

You can get started quickly.

1. Deployment observability component on your Kubernetes cluster.
2. Configure your value.yaml to enable observability features.
3. Apply values.yaml according to helm install or kubectl command.
4. (Optional) Deploy observability tools like Grafana on your Kubernetes Cluster.

When the deployment is successful, you can monitor metrics, logs, and traces data on your deployed observability tools. For more details, please refer to the observability configuration document.



A highly scalable distributed fast approximate nearest neighbor dense vector search engine.