StreamingFast February 2026 Update
Hey everyone — here’s a look at what we shipped in February across Substreams, Firehose, and supporting infrastructure, plus what’s currently in progress.
This month focused heavily on performance improvements, transport-layer optimizations, and preparing for upcoming ecosystem shifts.
What we shipped
Substreams SQL Sink: Speed Improvements + Delta Updates – The latest release of substreams-sql-sink introduces meaningful speed improvements thanks to usage of a new compression algorithm (S2). We also officially released Delta Updates support, making incremental aggregations easier and more efficient. Developers can now maintain rolling metrics (volume, counts, sums, min/max, etc.) using atomic, incremental updates without reprocessing full tables. See last month’s update for a bit more context if you’re interested.
BlockScopedData Batching Fully Deployed – Batching of BlockScopedData is now fully deployed, improving throughput and reducing overhead across streaming workloads.
Substreams Rust SDK: Quick Protobuf Support – Updated the Rust SDK to support Quick Protobuf, reducing encoding/decoding overhead and improving overall performance characteristics for Substreams modules.
What’s in flight
Latency Benchmarking & Optimization – Ongoing benchmarking work continues to surface new optimization opportunities. While the effort is not yet complete, early results are promising and showing measurable improvements.
HTTP3 / QUIC Block Stream Protocol – Work is underway on a new HTTP3/QUIC-based block streaming protocol (reader/relayer). Early testing shows strong performance improvements, particularly for sparse data processing. This work is expected to further reduce latency and improve reliability once production-ready.
op-reth Work Resumed – Development work has resumed on op-reth, following Base’s announcement that they are moving off the OP Stack toward their own stack. We’re adapting accordingly to ensure continued compatibility and long-term support.
Flashblocks for Base: Rewrite Required – Due to Base’s architectural changes, Flashblocks support for Base will need to be rewritten. As a result, removal of the Beta tag may take longer than originally anticipated.
Stellar Full Integration (Under Evaluation) – We are evaluating the possibility of upgrading from the current RPC Poller model to a full integration for Stellar to improve reliability and performance.