Technical Note #2: On the Temporal Structure of Distributed Pipelines

Technical Note #2 in a series on a distributed real-time market-data system.

This note proposes a change of viewpoint for studying timing effects in distributed pipelines. It treats a pipeline as a transformation from input time to output time, so that delay levels, local stretching or compression, and persistent timing episodes can be analyzed as parts of the same phenomenon.

The framework is applied to approximately 280 million events from the real-time market-data pipeline introduced in Technical Note #1. The case study shows how this viewpoint can be used on a live system and why it matters: the temporal transformations induced by the pipeline can be linked to changes in the interpretation of a downstream multi-exchange signal.