An Interactive Metrics Dashboard for the Keck Observatory Archive

TL;DR

This paper presents a new interactive dashboard for the Keck Observatory Archive that provides real-time metrics on archive performance and growth, supporting scalability and future planning.

Contribution

The paper introduces a scalable, Python-based, VO-compliant query infrastructure and a live metrics dashboard to improve archive monitoring and management.

Findings

Query speed increased by a factor of 20

Real-time data ingestion within one minute of creation

Enhanced monitoring capabilities for archive health

Abstract

Since 2004, the Keck Observatory Archive (KOA) has operated as a NASA-funded collaboration between the NASA Exoplanet Science Institute ( NExScI) and the W.M. Keck Observatory. It ingests and serves all data acquired by the twin 10-meter Keck telescopes on Mauna Kea, Hawaii. In the past three years, KOA has begun a modernization program to replace the architecture and systems used since the archive's creation with a new modern Python-based infrastructure. This infrastructure will position KOA to respond to the rapid growth of new and complex data sets that will be acquired by new instruments now in development, and enable follow-up to identify the deluge of alerts of transient sources expected by new survey telescopes such as the Vera C. Rubin Observatory. Since 2022, KOA has ingested new data in near-real time, generally within one minute of creation, and has made them immediately accessible to observers through a dedicated web interface. The archive is now deploying a new, scalable, Python-based, VO-compliant query infrastructure built with the Plotly-Dash framework and R-tree indices to speed-up queries by a factor of 20. The project described here exploits the new query infrastructure to develop a dashboard that will return live metrics on the performance and growth of the archive. These metrics assess the current health of the archive and guide planning future hardware and software upgrades. This single dashboard will enable, for example, monitoring of real-time ingestion, as well as studying the long-term growth of the archive. Current methods of gathering metrics that have been in place since the archive opened will not support the archive as it continues to scale. These methods suffer from high latency, are not optimized for on-demand metrics, are scattered among various tools, and are cumbersome to use.