Case Study: PhonePe scales 2,000% and saves $5M/year with Acceldata

How PhonePe scaled by 2000% and saved $5 million with Acceldata

PhonePe, a Walmart subsidiary serving more than 350 million consumers across India, runs a high-volume, real-time payments platform built on open-source technologies like HBase, Hive, Kafka, Spark and HDFS. As usage exploded, PhonePe faced major scaling, performance and visibility challenges in its Hadoop-based data infrastructure and engaged Acceldata’s data observability platform to gain better operational insight and reliability.

Acceldata was implemented in under 24 hours to provide real-time monitoring, automated alerts and root-cause analysis across HBase, Spark and Kafka, distinguishing hardware or table issues from seasonal surges. Within 18 months PhonePe scaled from 70 to 1,500+ nodes (~2,000% growth), achieved 99.97% Hadoop availability, reduced processing time by 55%, eliminated Sev‑1 incidents for 18 months, freed engineers from firefighting, and saved about $5 million per year in software licensing.

PhonePe

Burzin Engineer

Founder and Chief Reliability Officer