Summary
- The updated Leios visualizer now supports detailed scenario analysis and replay
- Networking experiments uncovered latency caused by buffer bloat, enabling more accurate testing with Linux traffic control
- New observability tooling with Grafana provides clearer insights into network behavior and prototype performance
- Research continued on optimistic performance, focusing on mempool consistency and MEV dynamics
- The new public Leios development tracker launched to give the ecosystem real-time visibility into progress.
On the November Leios monthly review and demo call, the team highlighted progress on grounding theoretical expectations in real measurements, improving tooling, and strengthening community visibility as Leios advances toward production readiness.
A refreshed visualizer for clearer insights
IOG’s technical architect, Sebastian Nagel, opened the session with a live demonstration of the updated Leios visualizer. Originally designed for large-scale scenarios with thousands of nodes, it has now been refreshed to support detailed, small-scale analysis that better reflects ongoing prototype testing.
The tool now allows users to:
- Replay recorded network scenarios step-by-step
- Slow down, accelerate, or time-travel through event timelines
- Inspect block and transaction propagation visually
- Observe transaction loads approaching the Leios stretch goal of 1,000 transactions per second (TPS).
This enhanced visualizer offers developers a clearer way to understand diffusion behavior, investigate anomalies, and contrast simulation output with real measurements as prototypes evolve.
Networking prototype: investigating real-world latency
A major focus of November was understanding the unexpected three-second latency discovered in October’s networking tests. The team used a three-node setup consisting of:
- `U node` (upstream): a lightweight sender providing extended blocks
- `node 0`: a patched Cardano node running early Leios mini-protocols
- `D node` (downstream): receiving and relaying data.
Identifying the anomaly
Through detailed logs, the team found that `node 0` requested 5 MB of data immediately after learning about a new endorser block (expected behavior). However, the upstream node managed to write the entire 5 MB nearly instantly. This contradicted the intended bandwidth limitations of the test harness.
Why the original setup failed
The team discovered that the original setup, which was based on Toxiproxy, did not accurately simulate transmission control protocol (TCP) behavior. As a result, data was able to move much faster than intended. By switching to Linux traffic control, the team was able to create more realistic network conditions, which significantly improved the accuracy of the measurements. This work confirmed that buffer bloat (where large data transfers delay smaller, time-sensitive messages) can impact block propagation and needs to be addressed as Leios traffic increases. Future tests will involve running on actual machines across real Internet paths.
New observability tooling for Leios experiments
To support thorough analysis, IOG’s engineer Dražen Popović introduced a new observability stack that brings together metrics, logs, and socket-level insights into a unified view.
Using process compose and a Grafana-based setup, the team can now:
- View detailed TCP socket statistics
- Visualize key Cardano node metrics
- Examine consolidated logs from consensus, networking, and storage components
- Use custom dashboards built specifically for Leios experiments, including views dedicated to buffer bloat and transport latency.
This tool enables engineers to move from raw logs to actionable insights and will be critical as testing becomes more complex.
Advancing research on optimistic performanceIOG’s engineer Giorgos Panagiotakos provided updates on the theoretical analysis of the optimistic Leios performance, specifically detailing the conditions under which the protocol can achieve low latency and high throughput.
Current areas of focus include:
- Mempool consistency: examining how similar transaction sets across stake pool operators impact the speed of block diffusion
- Geographical distribution: investigating how intercontinental latencies affect mempool divergence during periods of high load
- MEV dynamics: analyzing how different Miner Extractable Value (MEV) patterns may influence fork behavior or operator incentives, particularly concerning the distinction between ranking blocks and endorsement blocks.
These insights help align real-world expectations with the protocol's guarantees and inform ongoing refinements to the Leios design document and Cardano Improvement Proposals (CIPs).
Quality assurance and benchmarking foundations
Michael Karg, Cardano’s performance and tracing lead, outlined continued work on testing and protocol validation, which includes:
- Establishing a formal definition of trace evidence for Praos and Leios
- Exploring linear temporal logic (LTL) for lightweight conformance testing
- Laying the early groundwork for long-running benchmarks on dedicated hardware to ensure reproducibility and long-term comparability.
These activities are essential for strengthening the foundation for reliable, measurable progress as the Leios implementation expands.
Transparency through the new development tracker
Amani El Sheikh, a member of the Leios product team, unveiled the new public Leios development tracker. This new tool provides the ecosystem with:
- Real-time visibility into commits and ticket updates
- A clear mapping of work to product objectives and milestones
- Links to documentation, repositories, and recent updates.
This initiative significantly strengthens transparency for SPOs, developers, researchers, and community members following Leios development.
Additionally, the refreshed Leios documentation site also went live, featuring an updated design and direct access to the roadmap, technical documents, and testing updates.
Join the next call
The November call demonstrated steady progress across research, networking, testing, and public transparency. As the team continues refining measurements and expanding prototypes, the community is invited to follow along.
Leios monthly review and demo calls are open to everyone and streamed live on YouTube. Join us to hear the latest updates and share your questions.
