Case Study Deskcasestudydesk.com
AdTechSourced

Case Study: PubMatic cuts HDFS block footprint 30% and saves millions in licensing with Acceldata

PubMatic Case StudySourced & dated by Case Study Desk
Key facts · TL;DR
Company
PubMatic
Industry
AdTech
Challenge
Excessive MTTR and high infrastructure and OEM support costs at hyperscale.
Headline result
PubMatic reduced its HDFS block footprint by 30% and saved millions of dollars a year in software licenses across a 3,000+ node, 150+ PB Hadoop environment using Acceldata.

Key results

30%
HDFS block footprint reduction
Millions/yr
Software license savings
reduced OEM support costs

The challenge

PubMatic, one of the largest U.S. AdTech companies, serving roughly 200 billion daily ad impressions and processing more than 2 petabytes of new data per day, experienced excessive MTTR, frequent outages, and performance bottlenecks driven by its massive scale, including as many as 1,500 nodes in a single cluster. It also carried high infrastructure and OEM support costs it wanted to reduce.

The solution

The Acceldata platform isolated bottlenecks, automated performance improvements, and distinguished between mandatory and unnecessary data to scale PubMatic's big-data environment reliably. This correlated events across infrastructure, data layers, and pipelines to support its mission-critical and customer-facing analytics.

The results, in context

HDFS optimization reduced PubMatic's block footprint by 30%, Kafka cluster consolidation saved infrastructure costs, and reduced OEM support costs saved millions of dollars per year in software licenses. Acceldata also eliminated day-to-day engineering firefighting on outages and performance issues, letting data teams focus on growing the business.

Products used

Acceldata Acceldata Data Observability Platform