ReleaseBytes
aws AWS Big Data Blog ·

How Moovit Achieved 33% Cost Optimization Through AWS Architectural Modernization

blogdataawsarchitecthealthcareaws-redshift
announcement

Moovit, a leading Mobility-as-a-Service provider, optimized its data platform architecture on AWS, achieving 33% cost reduction and a 50% decrease in its Amazon Redshift cluster size. The company transitioned from a single-engine data platform to a multi-engine lakehouse architecture, distributing workloads across provisioned Amazon Redshift, Amazon Redshift Serverless, and Amazon EMR. This modernization involved systematically gaining workload visibility, cleaning up unnecessary processes, and selecting appropriate workloads for offloading. The approach ensures reliability and cost-efficiency for their growing data demands, supporting product analytics, BI, monitoring, and data science.

  • →Addressing Data Platform Scalability Challenges
  • →Gaining Workload Visibility with Automated Query Attribution
  • →Reducing Unnecessary Data Warehouse Load
  • →Adopting a Multi-Engine Lakehouse Architecture
Notes (4) ›
  • Addressing Data Platform Scalability Challenges

    Moovit's Amazon Redshift cluster became the backbone of their data platform, processing vast volumes of mobility and operational data for diverse workloads like ETL, BI, and data science. As storage and usage grew, the single-engine setup faced increasing queue times and SLA risks, necessitating a more scalable and cost-efficient architecture.

  • Gaining Workload Visibility with Automated Query Attribution

    A critical first step was implementing automated query attribution to classify each query by workload owner, execution context, and resource consumption. This provided a historical map of platform usage, enabling evidence-based decisions for optimizing or offloading workloads, prioritizing those with the largest and most stable optimization opportunities.

  • Reducing Unnecessary Data Warehouse Load

    Before architectural changes, Moovit identified and removed unused processes, reduced unnecessary job frequencies, and reviewed workload management guardrails. This cleanup phase eliminated around 7% of overall CPU time, ensuring that only genuinely heavy and necessary workloads were considered for offloading.

  • Adopting a Multi-Engine Lakehouse Architecture

    Moovit re-architected their data platform to a multi-engine lakehouse, combining provisioned Amazon Redshift, Amazon Redshift Serverless, and Amazon EMR. This allowed them to assign each workload to the most suitable engine, reducing pressure on their main Redshift cluster and creating a flexible, cost-optimized platform for future use cases.

Read the original announcement →

https://aws.amazon.com/blogs/big-data/how-moovit-achieved-33-cost-optimization-through-architectural-modernization/

Related releases