Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Beyond Performance: Measuring the Environmental Impact of Analytical Databases

Michail Bachras, Hans-Arno Jacobsen

TL;DR

This paper tackles the environmental footprint of analytical databases by introducing ATLAS, a systematic methodology that jointly accounts for operational and embodied emissions as well as water footprints. It pairs EcoQuery, a measurement infrastructure, with ATLAS to quantify how architectural choices (DuckDB, Hyper, MonetDB, StarRocks), hardware configurations, and deployment geography shape environmental impact. The study demonstrates that deployment location can dominate environmental differences due to regional energy mixes and water use in power generation, and it highlights trade-offs across storage media and hardware lifecycles. Practically, the work enables environmentally informed database design and deployment decisions, including break-even analyses and lifetime considerations for hardware investments. The integration of environmental metrics into database evaluation represents a significant step toward greener data systems with tangible guidance for developers and operators.

Abstract

The exponential growth of data is making query processing increasingly critical for modern computing infrastructure, yet the environmental impact of database operations remains poorly understood and largely overlooked. This paper presents ATLAS, a comprehensive methodology for measuring and quantifying the environmental footprint of analytical database systems, considering both operational impacts and manufacturing costs of hardware components. Through extensive empirical evaluation of four distinct database architectures (DuckDB, MonetDB, Hyper, and StarRocks), we uncover how fundamental architectural decisions affect environmental efficiency. Our findings reveal that environmental considerations in database operations are multifaceted, encompassing both immediate operational impacts and long-term sustainability implications. We demonstrate that architectural choices can significantly influence both power consumption and environmental sustainability, while deployment location emerges as a critical factor that can amplify or diminish these architectural advantages.

Beyond Performance: Measuring the Environmental Impact of Analytical Databases

TL;DR

This paper tackles the environmental footprint of analytical databases by introducing ATLAS, a systematic methodology that jointly accounts for operational and embodied emissions as well as water footprints. It pairs EcoQuery, a measurement infrastructure, with ATLAS to quantify how architectural choices (DuckDB, Hyper, MonetDB, StarRocks), hardware configurations, and deployment geography shape environmental impact. The study demonstrates that deployment location can dominate environmental differences due to regional energy mixes and water use in power generation, and it highlights trade-offs across storage media and hardware lifecycles. Practically, the work enables environmentally informed database design and deployment decisions, including break-even analyses and lifetime considerations for hardware investments. The integration of environmental metrics into database evaluation represents a significant step toward greener data systems with tangible guidance for developers and operators.

Abstract

The exponential growth of data is making query processing increasingly critical for modern computing infrastructure, yet the environmental impact of database operations remains poorly understood and largely overlooked. This paper presents ATLAS, a comprehensive methodology for measuring and quantifying the environmental footprint of analytical database systems, considering both operational impacts and manufacturing costs of hardware components. Through extensive empirical evaluation of four distinct database architectures (DuckDB, MonetDB, Hyper, and StarRocks), we uncover how fundamental architectural decisions affect environmental efficiency. Our findings reveal that environmental considerations in database operations are multifaceted, encompassing both immediate operational impacts and long-term sustainability implications. We demonstrate that architectural choices can significantly influence both power consumption and environmental sustainability, while deployment location emerges as a critical factor that can amplify or diminish these architectural advantages.
Paper Structure (40 sections, 2 equations, 22 figures, 4 tables)

This paper contains 40 sections, 2 equations, 22 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Background
  4. Database systems
  5. DuckDB
  6. Hyper
  7. MonetDB
  8. StarRocks
  9. Energy Profiler
  10. Hardware Setup
  11. Environmental Metrics
  12. Operational Carbon Emissions
  13. Embodied Carbon Emissions
  14. CPU Carbon Emissions Calculation
  15. Operational Water Footprint
  16. ...and 25 more sections

Figures (22)

  • Figure : (a) CPU and DRAM carbon emissions for TPC-H queries
  • Figure : (a) CPU carbon emissions for TPC-H queries
  • Figure : (a) Hyper carbon emissions across 2022 in different locations
  • Figure : (a) CPU Carbon Emissions by storage medium and DBMS
  • Figure : (a) Number of queries until operational and embodied carbon emissions are equal
  • ...and 17 more figures