Alertissimo -- a tool for orchestration of LSST broker streams

TL;DR

Rubin/LSST will generate a high-volume alert stream that must be ingested and processed by multiple brokers. The authors propose Alertissimo, a DSL-driven workflow engine with an input-agnostic core that translates user intents into intermediate representations and orchestrates calls to varied broker APIs via an Abstract Broker interface. Key contributions include the architecture for end-to-end processing, the DSL for Transients with a demo branch, and a prototype demonstrating multi-broker interoperability with planned NLP/visual interfaces. The work enables scalable, cross-broker pipelines for alert handling in the LSST era, potentially widening access and reducing orchestration friction for astronomy communities.

Abstract

The Vera C. Rubin Observatory, through its Legacy Survey of Space and Time, will soon start producing 10 million alerts on transient astronomical objects per night. Due to logistics and bandwidth, alerts will not be dispatched directly to the public but to 'brokers' i.e. tools selected by LSST to handle alert streams. Brokers offer both common, specific and micro-specific functionalities related to alert handling, analysis, representation and dissemination. In this ecosystem, potentially augmented by data streams from other astronomical sources, there is a - need demonstrated by the community - for use cases which combine features of individual brokers. In this paper we present initial efforts and a prototype of such a tool, along with a language that would allow users to define use cases / workflows in a manner tailored for the domain.

Figures (1)

• Figure 1: End to end flow diagram of Alertissimo modules illustrating the pipeline from DSL input through orchestration to broker execution. For detailed description see Subsection \ref{['architecture']}