Building Real-Time Analytics Systems

Chapter 3. Introducing All About That Dough: Real-Time Analytics on Pizza

In this chapter we’ll introduce a pizza delivery service called All About That Dough (AATD) that specializes in pizzas with Indian toppings. It’s been operating for almost two decades and has built up an infrastructure that allows the company to serve thousands of page views per minute on its websites, with a peak of 50 orders per minute.

AATD deals with three types of data: products, users, and orders. When it started out, the company used a relational database to store all this data, but a couple of years ago it introduced a streaming platform to deal with orders.

Every hour the data processed by the streaming platform is batched up before being loaded into a data warehouse, where analytical queries are run each night to see what’s happened in the business that day. AATD also store access logs of all visits to the website but so far hasn’t done anything with this data.

While AATD has done reasonably well at growing its business, several recurring problems are prohibiting further improvement:

• The company sometimes receives fraudulent or prank orders but doesn’t realize it until order processing has begun or even sometimes until days after the orders have been processed.

• Inventory management is difficult because the company doesn’t have a good idea of which products are popular at any particular moment. On multiple occasions, the company ran out of ingredients because of a surge in particular lines of ...