Grafana released Tempo 2.8 on June 12, 2025, introducing substantial memory optimizations and expanded functionality in its trace query language, TraceQL. This update is part of an ongoing effort to make distributed tracing more performant and accessible within observability stacks.
The most notable improvement is a greater than 50% reduction in peak memory consumption for the Tempo compactor. By profiling with Pyroscope flame graphs, Grafana engineers traced high memory usage to aggressive pooling. Replacing it with lighter pooling strategies and leveraging Go’s garbage collector led to significantly lower memory pressure, particularly under high-throughput workloads. This improvement directly benefits teams operating large-scale, latency-sensitive systems by reducing infrastructure costs and improving stability.
On the query side, TraceQL gains several new features. A new most_recent=true hint enables users to retrieve the latest traces in a deterministic way, which is particularly useful for debugging or identifying recent anomalies. Support for span:parentID filters adds more power to hierarchical trace analysis, allowing users to understand causal relationships in complex request chains. New metric functions like sum_over_time, topk, and bottomk expand Tempo’s analytical capabilities, allowing teams to identify performance bottlenecks or underutilized paths across their trace datasets.
Operational changes include a safer default HTTP port (3200 instead of 80), concurrent flushing for faster ingestion, iterator performance improvements, and tighter constraints on attribute sizes. Additionally, Tempo 2.8 includes stricter security defaults, with support for Go 1.24 and the use of distroless container images to reduce potential attack surfaces.
Community response has been broadly positive. Grafana’s X account stated the release brings memory improvements, new TraceQL features, bug fixes, and some breaking changes, while Deutsche Bank’s Florin Lungu commented on LinkedIn that Tempo 2.8 showcases Grafana’s commitment to optimizing user experience through performance gains and enhanced query capabilities. Grafana team member Mike McGovern highlighted new TraceQL functions, memory optimizations, and updated defaults in his post:
This update packs powerful TraceQL enhancements, significant memory optimizations, and smart configuration updates to improve performance and usability.
As organizations continue scaling observability infrastructure, Tempo 2.8 presents a compelling update, offering leaner performance, more expressive trace queries, and improved defaults out of the box. The full changelog and upgrade guidance are available on the official blog post and release notes.