Reliability & Operations
Designing for the day it breaks — targets, containment, and what you can see from the outside.
10 topics
Availability, SLOs & Error Budgets
Turning "it should be reliable" into a number, and using the gap between that number and 100% as a budget you're allowed to spend.
core20 minFailure Modes & Blast Radius
Asking "what breaks when this breaks?" for every box in the design, and containing the answer before it becomes an outage.
core20 minTimeouts, Retries & Backoff
The three settings that decide whether a dependency's bad minute is invisible or takes your service down with it.
core20 minCircuit Breakers & Bulkheads
Two patterns for stopping one broken dependency from consuming your service — fail fast when it's down, and cap what it can consume when it's slow.
core20 minGraceful Degradation & Load Shedding
Deciding in advance which parts of the product are allowed to disappear, so overload costs you a feature instead of the whole page.
core20 minObservability at Design Altitude
What you'd instrument and alert on, decided while drawing the system rather than after the first incident.
core20 minRollouts & Safe Deploys
Shipping changes to a running system without downtime — canaries, flags, and why every migration is two deploys.
core20 minCapacity Planning & Autoscaling
Knowing how much you can serve before you need to, and the reasons autoscaling doesn't save you from finding out.
core20 minError Handling
Deciding which failures are expected and handled, which are bugs that should crash loudly, and who finds out either way.
core25 minsharedObservability Fundamentals
Instrumenting a system so you can answer questions about it you didn't think to ask before it broke.
core25 minshared