Skip to content
Ameen Abozaid
MID-049

Reliability

  • 6 Lessons
  • 2 h practice (estimate)
On this page 8 topics

Overview

Topics

  • SLIs
  • SLOs
  • error budgets
  • availability
  • latency percentiles

Learn first

2 h of practice (estimate)

Learning outcomes

  1. Define SLIs that reflect user experience (availability, latency percentiles, correctness).
  2. Set SLOs and compute error budgets.
  3. Alert on error-budget burn rate.
  4. Explain how error budgets guide release and reliability decisions.

Lessons

  1. 01What reliability means to users
  2. 02SLIs
  3. 03SLOs and error budgets
  4. 04Burn-rate alerts
  5. 05Using the budget
  6. 06Demo: SLO dashboard and burn-rate alert for checkout

Lab and exercises

Lab

MID-049-LAB1 · Checkout SLOs

Outcomes: 1–4 · Time: ~2 h Define SLIs and SLOs for checkout and search, with burn-rate alerts, and write a short SLO document including the budget policy. Acceptance: SLO document reviewed; alerts tested with injected errors.

Exercises (assessment: SLO document review)

IDOutcomeTask
MID-049-EX011For Shop's search and checkout, write one availability SLI and one latency SLI as good/valid ratios.
MID-049-EX021Explain why a SLI measured inside the service can miss failures users see.
MID-049-EX032Compute the error budget in requests and in minutes for a 99.9% SLO over 30 days with 10 million requests.
MID-049-EX042Propose an SLO for search and justify why it is not 100%.
MID-049-EX053Write a two-window burn-rate alert (page and ticket) for checkout.
MID-049-EX064Write a budget policy with actions at three remaining-budget levels.
MID-049-EX071–4Review a sample SLO document with a 100% target and server-side SLIs; list fixes.
Show solutions
  • EX01 Availability: successful checkouts / valid checkout requests; latency: search responses under 300 ms / all search responses.
  • EX02 Requests rejected at the load balancer or gateway, or timeouts before reaching the service, never appear in service metrics.
  • EX03 Budget 0.1%: 10,000 failed requests; about 43.2 minutes of full outage (30 days × 24 h × 60 × 0.001).
  • EX04 For example 99.5% of searches under 300 ms; users tolerate an occasional slow search; the cost of 100% is unbounded.
  • EX05 Page: 1h and 5m ratios above 14 × 0.001; ticket: 6h and 30m above 6 × 0.001 (thresholds are a team choice).
  • EX06 Above 50%: normal releases; 10-50%: extra review and canaries; below 10%: reliability work only until recovered.
  • EX07 Replace 100% with a realistic target, measure at the gateway, add the budget policy and burn-rate alerts.

Quiz

Outcomes: O1 SLIs · O2 SLOs and error budgets · O3 burn-rate alerts · O4 using the budget.

A self-check for you: nothing is graded, saved or sent anywhere. Choose an answer, then check it. You can try again as often as you like.

  1. 1An SLI is typically:O1
  2. 2Best place to measure availability for users:O1
  3. 3A latency SLI could be:O1
  4. 4The error budget of a 99.9% SLO is:O2
  5. 5Why not set a 100% SLO?O2
  6. 699.9% over 30 days allows roughly:O2
  7. 7Burn rate 1 means:O3
  8. 8Combining a long and a short window in a burn-rate alert:O3
  9. 9When budget is nearly spent, teams should:O4
  10. 10The budget policy should be:O4

Practical part

Seven exercises MID-049-EX01 to EX07 (EX07 is the SLO document review) plus the checkout SLO lab. Pass: at least five exercises and a lab with a reviewed SLO document and burn-rate alerts tested by injected errors.

Common mistakes

100% targets; SLIs measured at the server rather than as users experience them.

Assessment

SLO document review.

Career relevance

SLO engineering (SRC-012).

Type to search levels, phases, modules and lessons.