Dash0 introduces SIFT: an observability framework that uses AI to reduce, enrich, and better structure the amount of telemetry. This should lead to faster error analysis and lower costs.
At KubeCon, Dash0 launches an observability framework called SIFT. With this, the company aims to provide an answer to what it sees as a growing problem within observability: the increase in telemetry data.
While traditional tools mainly focus on collecting and storing more and more data, Dash0 opts for a different approach. SIFT filters irrelevant data, improves the remaining data with AI, and makes it easier to establish connections between alerts.
Acronym
SIFT is built around four pillars, from which it derives its name: Spam Removal, Improve Telemetry, Filtering & Grouping, and Triage. With Spam Removal, users can filter out irrelevant data with one click. Dash0 automatically generates OTTL rules that block unwanted information before it is stored. This, in principle, reduces storage costs and keeps the data stream manageable. The functionality is also available in Kubernetes via the Dash0 Operator.
For Improve Telemetry, Dash0 uses AI to automatically structure logs, apply semantic standards, and recognize patterns in logs and traces. This provides developers with more context for error messages and requires less manual standardization.
Automatic Error Analysis with AI
The third component, Filtering & Grouping, makes every interface component clickable and interactive. Users can see the effect of their filters in advance. The new Grouping functionality automatically groups similar data.
With the fourth pillar, Triage, SIFT performs an automated error analysis. The tool identifies the likely cause of a problem with one click and displays relevant metadata in a visual overview. Dash0 is working on an extension of Triage where AI agents independently test hypotheses, thus automatically identifying the cause of problems.
SIFT is available to all existing Dash0 customers. According to the company, the framework leads to shorter resolution times and lower costs, without compromising on insight. The solution is built on open standards such as OpenTelemetry and uses AI and ML technologies.