Case Study: PubMatic achieves massive performance optimization and millions in cost savings with Acceldata Pulse

PubMatic Optimizes Performance and Cost at Massive Scale

PubMatic, a recently public, fast-growing American ad-tech firm processing ~200 billion daily ad impressions and over 2 PB/day of new data on a hyperscale stack (HDFS, Yarn, Kafka, Spark, HBase), faced excessive MTTR, frequent performance issues from thousands of nodes in a single cluster, and high infrastructure and OEM support costs. To address these challenges they engaged Acceldata and used Acceldata Pulse for data observability and system tuning.

Acceldata Pulse isolated bottlenecks, automated performance improvements, and separated mandatory from unnecessary data to reliably scale PubMatic’s big data environment. As a result, HDFS block footprint was reduced by 30%, Kafka cluster consolidation cut infrastructure costs, OEM support/license costs were reduced by millions of dollars per year, and day‑to‑day engineering firefighting was eliminated so teams could focus on scaling the business.

PubMatic

Ashwin Prakash

Analytics Engineering Leader Pubmatic Data