A scheduled exam-start spike

A scheduled start needs one coordinated readiness plan and a rehearsal of the traffic that will actually arrive.

Before you read: predict

A scheduled catalog rehearsal recorded five effects: login admission retries occupied API slots; the first reservation failed when its proxy route was absent; a bundle metadata miss left the release screen blank; dashboard charts lagged; and the weekly email was delayed.

Which two investigation areas should lead?

The 30-second version

A scheduled spike needs a rehearsed launch plan. Map dependencies, compare their failure effects, record evidence from the release path, and use the observed arrival shape in the test. Keep the plan specific to this event rather than treating an average day as its substitute. S1

A scheduled start needs one coordinated readiness plan and a rehearsal of the traffic that will actually arrive.

How it works

Step 1 of 7

Record the known opening and expected arrival shape; scheduled capacity preparation can accompany responsive scaling. S2

Analogy: A guided cave tour with a fixed departure

Lighting checks, equipment preparation and visitor check-in can proceed independently. Before the group departs, the guide checks that the required preparations have finished. A holding area keeps visitors organised but cannot make an unfinished route ready. The useful object is the joint departure checklist, not a longer list of unrelated jobs.

Tour coordination	Launch coordination
Fixed departure	Published opening
Independent preparation teams	Fleet, pooler and paper owners
Joint readiness check	Evidence before admission
Rehearsing the group departure	Testing the concentrated launch workflow

In real terms: each preparation has an owner, dependencies and a completion signal. Breaks down when: this analogy concerns coordination only; it provides no cave-safety guidance. Contrasting case: a quiet rehearsal that walks through one person at a time may validate a route but miss a simultaneous departure. Similarly, an average-load test answers a different question from a scheduled rush. S4

There is no universal order between every preparation task. Independent work can run together; the join must wait for the prerequisites it actually consumes. A CDN-ready paper does not establish database readiness, and a healthy API process does not prove its intended pooler route was exercised. These are separate completion conditions in this authored runbook, not claims about any particular cloud product. S1

Try it

Answer first, then check. Nothing is saved.

Question 1 of 4

A release ledger records 3,600 arrivals at 09:00:00, 3,600 at 09:00:30, 900 at 09:01:00 and 900 at 09:02:00.

Which first-minute assertion should the rehearsal retain?

Question 2 of 4

At 08:52 the release record shows the fleet and the proxy route ready; the bundle is uploaded but its gate check has not run. Warmup verification is scheduled for 08:53 and admission for 09:00.

Which runbook change should the reviewer ask for?

Question 3 of 4

The runbook lists an API-group action and a proxy-group action at 09:00. One author treats the matching timestamps as a dependency order; another treats proxy creation order as that order.

Which review claim is supported?

Question 4 of 4

Sale telemetry shows that session refresh and reservation traffic account for 91% of event requests, while exports and weekly reporting each stay below 1%. A normal-day replay stays near the average load.

Which candidate claim best defines the spike test?

Go deeper

Pro tip

Write one owner beside each launch action before rehearsal. It turns a handoff into a conversation with someone accountable, instead of a checklist everyone assumes another team will finish. Google SRE, Rollout Planning, read 2026-09-30. This is a coordination habit, not an assessment option.

Checked against the tools and docs as of 2026-09-29.

Predict first