TraceQL Internals: Building a Query Language for Distributed Traces
Session Description
Distributed tracing is essential for understanding systems and service interactions in cloud-native systems.
l will discuss Tempo’s deep integration with OpenTelemetry, including how OpenTelemetry conventions, scopes, and specs shape its storage, query layer, and query language design.
This talk focuses on the design and implementation of TraceQL, why we built it, and deep dives into Grafana Tempo's storage and query layer and usages of object store to achieve low storage costs.
l will talk about how Grafana Tempo’s cost-efficient, scalable architecture leverages object stores and Apache Parquet to store trace data and power TraceQL, a purpose-built query language for trace data.
Key Takeaways
This talk revolves around Tempo’s architecture and how TraceQL helps you get more out of the Trace data at lower cost by making use of object stores and columnar storage format Apache Parquet.
Attendees will learn how and why behind the building TraceQL, and how it helps us make distributed tracing better and get more value out of the trace data.
We will deep dive into Tempo's Internals with a focus on the Query and Storage Layer.
References
Session Categories
Which track are you applying for?
Speakers
Suraj Sidh
Reviews
This is an interesting talk diving into the observability space.
I am marking it as not sure, as this talk is also very closely related to Grafana Tempo and TraceQL. Seems like it would fit well into Data devroom
This is a FOSS project and the proposer is one of the top contributors to the project. Observability talks are well attended at our events and this is one of the better proposals we've got this year from the domain. The proposal promises to be technical and focus on design/implementation but I would still to go for a mock to ensure this does not end up becoming a marketing talk.