New and updated features.

Distributed Tracing (Beta)

This release adds collection and visualization of distributed tracing data. We send trace data from your instrumented applications to the Wavefront service, and you can examine it with our new tracing UI. We augment the trace data with metrics and histograms to provide a full troubleshooting and visualization experience that starts with charts for application metrics and drills down to specific traces and spans to find bottlenecks.

Watch this video for an introduction to the Beta version. We’re continuously improving the UI and will post new videos over time.

introduction to tracing

Using traces, histograms and traces you can:

  • Monitor your application to performing as expected.
  • Troubleshoot and analyze reported errors.
  • Pinpoint the specific operations that bottlenecks occur in.

How you instrument your application for tracing depends on the tools and SDKs the application is using. Wavefront supports the OpenTracing standard.

Use our docs and the Beachshirts sample application to learn more.

tracing flow

Wavefront Observability SDKs

Wavefront has added a significant number of new SDKs that allow you to instrument your application stack at different levels. You can find our SDKs in our github repository. SDK releases and updates are completely independent from Wavefront service releases. We mention them in these release notes because other features in this release depend in part on the new SDKs.

Multi-Threshold Alerts

Classic alerts have a single severity level (e.g. SEVERE) and a set of alert targets. When the alert targets are notified, all alert targets get the notification. Multi-threshold alerts allow finer control:

  • When you create a multi-threshold alert, you associate different severity levels with different thresholds. For example, a CPU load over 5000 is SEVERE, a CPU load over 4000 is WARN.
  • You also associate one or more alert targets with each severity.

    Note: You can associate each alert target only with one severity. For example, suppose you set up an alert to notify target T1 for the SEVERE severity. You cannot specify target T1 for any other severity in that alert.

  • We continue to check the alert once a minute by default. For each affected severity we send alert notifications to each specified target. For example, if CPU load is over 4000 for the specified amount of time, the set of targets associated with WARN and lower severities are notified. Later, if CPU load is over 5000, the set of targets associated with SEVERE and lower severities are notified.

Creating a threshold alert is similar to creating a classic alert.

Watch this video to learn more:

threshold alerts