SLIs, SLOs, and SLAs

I just watched a Google Cloud Tech talk from 2019 called: DevOps Vs. SRE: Competing Standards or Friends. You can watch it on YouTube, HERE and if you want to skip to the part about SLIs, SLOs, and SLAs you can go HERE.

In this talk, Seth Vargo brings up three concepts that I think are key to understand in order to grasp DevOps and how monitoring plays a key role if you want to successfully develop and deploy software.

SLI

An SLI is a Service Level Indicator. At any point in time, an SLI will indicate whether or not a service is satisfying a specific metric. Examples brought up in the talk are: Request latency, requests per second, and failures per request.

If you have a requirement that the ratio between successful and failed requests towards a particular endpoint must always be above 99.99%, the SLI will tell you whether you fulfill this requirement – at a particular point in time.

SLO

An SLO is a Service Level Objective; a binding target for a collection of SLIs. The SLO answers the question: “How many failing SLIs can we allow over a particular period of time?”

SLA

An SLA is a Service Level Agreement, which is a business agreement between a customer and a service provider. Typically, an SLA is based on several SLOs.

Summary

SLIs drive SLOs which inform SLAs.