SRE Series — 02 SLIs, SLOs, and SLAs
Software Engineering Topic of the Day #44
Credit: Originally written and published by Billie Thompson on Armakuni’s internal Slack
SRE Series — 02 SLIs, SLOs, and SLAs
Site Reliability Engineering (SRE) has a focus on the reliability of your software systems. An important principle of this is that the customer is the ultimate arbiter of your system’s reliable.
Service Level Agreements (SLA) are set up to set a minimum standard for reliability with the customer. To make these agreements matter, there are typically some sort of consequences for not meeting them, such as refunds or credits to the affected customers.
Service Level Indicators (SLI) will then need to be identified for the critical user journeys you provide. SLIs are a a quantitative metric or measurement of user experience. In plain English, this is a metric that shows how happy your users are with your service, on a per-event basis.
In order to meet our SLAs with the customer, we want to make sure our own targets for reliability are slightly more ambitious. We call these targets the Service Level Objectives (SLO). SLOs are challenging, but accomplishable goal written in terms of the SLI that are internal to your company.
It is important that SLOs are reviewed regularly, asking if they are adequate (or too ambitious) and whether the SLIs in fact correlate with customer happiness.
Q1: What are some metrics that you can monitor in your systems that correlate with customer happiness?
