Implementing Service Level Objectives

Chapter 11. Data Reliability

Polina Giralt and Blake Bisset

The opening chapters of this book discuss how we live in a world of services. We also live in a world of data. Most services create, collect, process, or present data in some way. We’re surrounded by data services! The goal of this chapter is to explore what makes SLOs for data services different from SLOs for other services.

First, we’ll define data services and consider our data users. The heart of this chapter covers measuring service objectives via 13 data properties. For each property, we’ll explore its measurement and its relationship to system design. We’ll finish with a short explanation of how to ensure data quality via service level objectives, to keep users happy.

Data Services

Welcome to the world of data service reliability. We’re bombarded with data each day. Financial data. Social data. Training data for algorithms. Data that is historical or near real time, structured or unstructured. Privately guarded corporate secrets as well as publicly available government datasets. Microservices consuming tiny amounts of JSON data from queues. Monolithic banking applications creating thousands of regulatory reports. And, of course, every other abstraction through which humanity has struggled to describe and make sense of the world since Grace Hopper plucked the first actual bug out of a computer.

Data application owners need to ensure that their services are reliable—but the essentials of data reliability ...