Know It Cold

This is not a lesson. Every domain page on this site teaches why a deployment strategy or a DR pattern works the way it does — this page skips the why and drills the numbers themselves, the ones a multiple-choice question drops on you cold with no context and expects back in under thirty seconds: the six domain weights, the 75/180/750 format numbers, the name and one-line tradeoff of every deployment strategy including the exact CodeDeploy config identifiers, and RTO/RPO defined precisely enough to survive a trick question. If you can recite this whole page from memory with the tab closed, that part of the exam is no longer a knowledge problem — it's just reading comprehension.

The format numbers — say them without thinking

☺ Like you're 10: Seventy-five questions, three hours, and you need a made-up score of 750 out of a possible 1000 to pass — and ten of those questions secretly don't even count.

These six numbers are the shape of the exam itself. Get them so automatic that reading a question never has to compete with wondering how much time you have left.

The six domain weights, in order

☺ Like you're 10: One domain is worth almost a quarter of the test on its own, three domains tie for a close second, and only one domain — the smallest — is worth less than a seventh of your score.

Six domains, six percentages, and they should come back to you in order without hesitating. Notice the tie: D2 and D6 are both 17%, and D3 and D4 are both 15% — a detail multiple-choice questions like to probe ("which is heavier weighted, Config Management or Security?" — trick question, they're identical).

Collapse it into a study-time mnemonic: D1 alone (22%) beats D4+D5 combined (29% vs. 22%... barely) — more precisely, D1+D2+D6 = 56%, over half the exam sitting in three domains, which is why a study plan that splits time evenly across all six is already wrong before it starts. See the DOP-C02 study plan for the week-by-week version of that math.

Every deployment strategy — name and the one-line tradeoff

☺ Like you're 10: Seven names, and each one trades speed of rollback against how much of production is exposed to a bad build at once — memorize the trade, not just the name.

Full explanations live in deployment strategies; this table is the compressed version for recall under pressure.

Two pairs the exam likes to blur: canary vs. A/B testing both split traffic by percentage, but a canary is a reliability signal torn down once verified, while an A/B test is a business experiment kept running on purpose. And rolling vs. recreate both use one environment, but only rolling keeps old and new serving traffic side by side — recreate accepts a downtime window specifically to avoid that.

The CodeDeploy configs, by exact name

☺ Like you're 10: AWS gives each rollout speed a literal name, and multiple-choice answers are picky about the exact spelling — "AllAtOnce" is not "OneAtATime" and the exam wants you to notice the difference on sight.

Vendor-neutral strategy names above map to CodeDeploy's own vocabulary differently depending on the compute target. Know both halves cold — this is the part of D1 (SDLC Automation) that's pure memorization, not reasoning.

# deployment group config — literal identifiers, exam-testable as strings
# EC2 / on-premises, in-place:
DeploymentConfigName: CodeDeployDefault.OneAtATime     # most fault-tolerant, slowest
# DeploymentConfigName: CodeDeployDefault.HalfAtATime
# DeploymentConfigName: CodeDeployDefault.AllAtOnce    # zero fault tolerance, fastest

# Lambda / ECS, traffic-shifting:
DeploymentConfigName: CodeDeployDefault.LambdaCanary10Percent5Minutes
# DeploymentConfigName: CodeDeployDefault.LambdaLinear10PercentEvery1Minute
# DeploymentConfigName: CodeDeployDefault.LambdaAllAtOnce

Three facts worth their own flashcard: ① Lambda traffic always shifts between two versions behind one weighted alias — never against raw version numbers, never against $LATEST. ② Linear keeps adding a fixed percentage forever until 100% — it has no permanent hold. Canary jumps once, holds once, then sends the rest — don't swap those two descriptions on a question. ③ Automatic rollback on either EC2 or Lambda/ECS fires only when a CloudWatch alarm is explicitly attached to that specific deployment group — an alarm that exists elsewhere in the account, unattached, does nothing. Full walkthrough in SDLC Automation and the AWS Developer Tools reference.

RTO vs. RPO, and the four DR patterns cold

☺ Like you're 10: RTO is the clock counting forward from the crash until you're back up; RPO is the clock counting backward from the crash to your last good copy of the data — and the four DR patterns are just four price tags for shrinking both clocks.

Two definitions, memorized so precisely you could recite them half-asleep:

And the four canonical DR patterns, cost and speed rising together in one fixed order — B-P-W-M: Backup & restore, Pilot light, Warm standby, Multi-site active/active.

One more cold fact that rides along with DR questions: lowering a Route 53 record's TTL does not make failover instant. Route 53 answers differently the moment it observes a failed health check, but resolvers and clients holding a cached answer honor it until that cache expires — a low TTL (60 seconds or less, set well before the incident) shrinks the failover window, nothing shrinks it to zero. Full derivation in Resilient Cloud Solutions.

The rest of the numbers worth having cold

☺ Like you're 10: A handful of other numbers show up again and again — the "best" delivery speed and failure rate a team can hit, and a couple of AWS port/behavior facts that questions love to test in passing.

Not the four headline topics above, but frequent enough on this platform's mock exams to earn a place on the same crib sheet:

Drill it — cover the page, write it blind

☺ Like you're 10: Reading this page again feels productive and mostly isn't — the version of studying that actually works is closing it and writing the numbers from nothing.

Recognizing a number when you see it and producing it from nothing are different skills, and only the second one is tested at minute forty under a countdown clock.

Run it more than once across your last week. The first pass tells you what's missing; the second tells you what didn't actually stick the first time. By the third pass these numbers should come out before you've finished reading the question — which is the only speed that's actually useful with a 180-minute clock and 75 questions splitting it 2.4 minutes apiece.

That's the crib sheet — format numbers, domain weights, every deployment strategy and its CodeDeploy name, RTO/RPO, and the stray facts that ride along with them. Pair it with No Docs Map — Closed Book for the reminder that none of this is available to look up mid-question, the AWS Service & Command Reference when you need the surrounding detail these tables deliberately strip out, and Answer Triage & Elimination for what to do when a question doesn't match anything on this page cleanly. Then take it into Mock Exam · Set 1 under a real 180-minute clock and see how much of this page you didn't actually need to stop and think about.