Site Reliability Engineering

Chapter 29. Dealing with Interrupts

Written by Dave O’Connor

Edited by Diane Bates

“Operational load,” when applied to complex systems, is the work that must be done to maintain the system in a functional state. For example, if you own a car, you (or someone you pay) always end up servicing it, putting gas in it, or doing other regular maintenance to keep it performing its function.

Any complex system is as imperfect as its creators. In managing the operational load created by these systems, remember that its creators are also imperfect machines.

Operational load, when applied to managing complex systems, takes many forms, some more obvious than others. The terminology may change, but operational load falls into three general categories: pages, tickets, and ongoing operational activities.

Pages concern production alerts and their fallout, and are triggered in response to production emergencies. They can sometimes be monotonous and recurring, requiring little thought. They can also be engaging and involve tactical in-depth thought. Pages always have an expected response time (SLO), which is sometimes measured in minutes.

Tickets concern customer requests that require you to take an action. Like pages, tickets can be either simple and boring, or require real thought. A simple ticket might request a code review for a config the team owns. A more complex ticket might entail a special or unusual request for help with a design or capacity plan. Tickets may also have an SLO, but response ...