"Pythian offered a wealth of Hadoop expertise to efficiently optimize, upgrade and maintain the Hadoop clusters that housed our mission-critical user data. If we had ANY questions, we could just can ask them! Their breadth of expertise and flexibility also meant they could also help us with our MySQL needs."

Overview

This web application provider relies on critical user data to continually improve the experience for their millions of global users. They turned to Pythian to help them make use of their Hadoop environment to effectively track user metrics on their central product that has millions of users worldwide.

Pythian Hadoop Services Optimize Mission-Critical User Data For A Global Web Application Provider

 

Critical Issues

Tracking user metrics is paramount for this company so they can better understand the uptake, prioritize fixes and new features of their main application—and ultimately improve the customer experience. However, their system for processing mission-critical user data and their key application web logs suffered from slow batch processing, so it took too long to see aggregate results. There was also no backup process, which meant a very real danger of losing vital user data. The answer was to move the process for tracking metrics and storing data in their Hadoop system—but the client lacked Hadoop expertise in house.
 

We Provided

The client engaged Pythian for its unparalleled expertise in supporting Hadoop environments. Pythian experts optimized, upgraded and maintained the Hadoop clusters that housed the company’s mission-critical user data. Improvements included enabling fast data recovery and right-sizing the clusters by reducing the number of Hadoop nodes from 30 to 20. These improvements to their Hadoop environment helped them improve the web log processing of user data in the the application’s health report, enabling them to easily access critical customer usage information, including crash reports and user locations. Pythian developers also anonymized the personally identifiable information (PII) that’s used to inform decisions in marketing, at the C level, and by R&D and the support desk. In the end, Pythian enabled the client’s internal team to continue managing the new environment  through an extensive knowledge transfer.
 

Result

The company CxOs, marketing, R&D and the support desk can now have confidence in the data they’re using to determine user trends and market share, prioritize features and fixes, and analyze campaign results using forensic metrics. Aggregate results are available much sooner than in the past, and a reliable backup process ensures that vital user data is secure. Reducing Hadoop nodes from 30 to 20 decreased hardware costs by one-third, providing savings of $100,000. And Pythian’s flexible service delivery model enabled the company to swap unused Hadoop hours for MySQL services, which freed up the company’s IT staff personnel for other tasks.
 

Technologies

  • Apache™ Hadoop®
  • Oracle®
  • MySQL
Share: