Case Studies
Company-wide operational analytics
Overview
Analytics & Modeling - Big Data Analytics Analytics & Modeling - Predictive Analytics Platform as a Service (PaaS) - Data Management Platforms | |
Professional Service Software | |
Business Operation Quality Assurance Sales & Marketing | |
Software Design & Engineering Services System Integration | |
Operational Impact
Performance was 10x or greater from the start, even without tuning. During the POC, the smaller join on Firebolt, which had been taking at least 40-60 seconds using Athena, was under 5 seconds. Even the larger join, which could not be done on Athena, ran within 5 seconds after a little optimization. Now even the larger joins can be run in 1-2 seconds, and filtering is millisecond. | |
When Appsflyer was using Athena, they could not analyze any data beyond 5 billion rows, which prevented them from performing a broad range of analyses. With Firebolt, there is no limit. Appsflyer is now able to analyze any level of detail, up to hundreds of billions of rows, with confidence. Firebolt’s ability to scale horizontally also meant it easily supported any number of concurrent users. | |
With Athena, cost had not been as much of a concern; it just didn’t work at scale. But it was a major concern as they were evaluating their alternatives, in part because of their scale. BigQuery ended up being the other option they considered. Beyond Firebolt’s faster performance, it also ended up being up to 10x lower cost, in part due to its greater efficiency per node. | |
Quantitative Benefit
Performance improvement of 10x or greater from the start. | |
Query times reduced from 40-60 seconds to under 5 seconds for smaller joins. | |
Larger joins that could not be done on Athena now run within 1-2 seconds. | |