Description

Shred packet streams have been redistributed between the networking and compute layers to minimize data-transfer latency. An intermediate buffering layer has been added, providing:

• pre-processing of shred packets before they reach BBM;

• filtering of duplicate data.

Shredstream now interacts directly with BBM, enabling:

• real-time delivery of up-to-date data for block construction;

• bidirectional synchronization of block and slot status;

• receipt of metadata for optimal assembly planning.

The interaction interface is implemented over the QUIC and gRPC protocols, which helps avoid redundant network calls and improves the reliability of data exchange.