Lion Cipher Lab
Work out a service's composite availability from its dependencies
- Cyber / Security Engineer
- Intermediate
- ~35 min
- Free · no account required
Why it matters
Availability targets are easy to promise and easy to miss. A request to a typical web service passes through DNS, a load balancer, web servers, application servers, a database, an identity provider and a payment API, and it fails if any one of them fails. The NIST/SEMATECH reliability handbook gives the arithmetic: components in series multiply their reliabilities, and redundant components in parallel fail only when all of them fail. Google's SRE book adds the scale that makes it matter: each additional nine is an order of magnitude improvement toward 100%. Teams regularly sign up for a…
What you'll learn
- series and parallel availability arithmetic
- turning availability into minutes of downtime a month
- finding the tier that contributes most downtime
- the ceiling set by dependencies you cannot replicate
- why redundancy in one failure domain is not redundancy
- checking an availability target before promising it
Steps in this lab
- Series multiplies; parallel only fails when everything fails
- The service stack
- Write the calculator
- Run it: the target was never reachable
- Use it before you promise a number