Creating a Data-Driven Enterprise with DataOps

Chapter 12. Capture All Data, Decide What to Do with It Later: My Experience at eBay

by Debashis Saha, former vice president of commerce platform infrastructure at eBay

Note

Debashis Saha is a former employee of eBay. The views expressed here are his own.

What can’t you buy on eBay? You name it, you can find it on the global online marketplace. Four nights in a luxury condo in Telluride? Credit for Uber rides? A Winnebago motor home? Or perhaps clothes, kitchen gadgets, computers, and just about anything else you can think of for personal or business use.

All the activities of all the millions of buyers and sellers—165 million active users at the end of 2016¹ to be exact, with more than 1 billion simultaneous listings²—are captured as part of eBay’s standard operating procedures: capture all data and decide whether—and how—to use it later.

When I joined eBay 10 years ago, I was glad to see that a data-driven culture was already ingrained in the organization. In particular, I was delighted to see that eBay already had executive support for centralizing data. It made my role to build, manage, and operate everything to do with data for eBay—from datacenters to frameworks, to data infrastructure, networks, and all data services—much easier.

At eBay, we produce three fundamental types of data. First is the transactional data: every time a user buys, bids, or otherwise takes action on the site, data is generated and collected. Transactional data produces tens of millions of data points ...