Skip to main content

Benchmarks

Performance Benchmarks

Real-world performance testing results for data processing workloads on Amazon EKS. TPC-DS benchmarks, Graviton comparisons, and acceleration technologies.

TPC-DS StandardReal DatasetsProduction Scale

Benchmark Results & Setup​

Benchmark Setup

Complete guide for setting up TPC-DS benchmark infrastructure, data generation, and test execution.

Data GenerationTest ConfigurationInfrastructure Setup
Setup Guide

EMR on EKS TPC-DS

Comprehensive TPC-DS 3TB benchmark comparing EMR on EKS performance across different configurations.

TPC-DS 3TBEMR on EKSCost Analysis
View Results

Spark Graviton R Series

Performance benchmarks comparing ARM-based Graviton processors with x86 instances for Spark workloads.

Graviton3R7g vs R6iCost/Performance
View Results

Spark Gluten + Velox

Acceleration benchmarks using Gluten and Velox vectorized execution engine for Spark on EKS.

Vectorized ExecutionNative Engine2-3x Speedup
View Results

About TPC-DS Benchmarks​

TPC-DS (Transaction Processing Performance Council - Decision Support) is the industry-standard benchmark for evaluating the performance of decision support systems. Our benchmarks use:

  • Dataset Sizes: 1TB and 3TB scales
  • Query Suite: 99 queries covering complex analytical patterns
  • Metrics: Query execution time, cost per query, resource utilization
  • Infrastructure: Amazon EKS with Karpenter autoscaling, Spot instances

All benchmarks are reproducible using the setup guides provided. Results include detailed methodology, infrastructure configuration, and cost analysis.