97 things every SRE should know

SRE is not a job, is a hat.

1. Measure with a goal in mind.

2. Analyze what you are measuring.

SLO -> Service level objectives | Set a “reliability” target for your service SLI -> Service level indicators | measure the system from the user point of view. Error budgets | How you performed on your SLOs

Is not your application is OUR application

Good-enough is good enough, aiming for perfection is expensive, both in human and financial resources.

Improve resilience:

1. Load reduction

   1. throttling

   2. Load shedding

   3. Prioritization

   4. queuing

   5. load balancing

2. Latency reduction

   1. Caching

   2. Regional replication

3. Load adaptation

   1. Autoscaling

   2. Over provisioning

4. Resilience

   1. Timeouts

   2. circuit breakers

   3. bulkheads?

   4. retries

   5. failovers

   6. failbacks

5. Meta techniques

   1. Improving tooling

   2. Scale faster (depending on situations)