Cloud Engineer
Cloud platforms (AWS, Azure, GCP) questions
170 questions on Cloud platforms (AWS, Azure, GCP) for cloud engineers, easiest first.
Easy 57
- Who patches the VM operating system?easy
- Regions and availability zoneseasy
- Choosing an S3 storage classeasy
- Horizontal vs vertical scalingeasy
- How serverless functions are billedeasy
- Usable IPs in a /24 subneteasy
- Shared file system for many serverseasy
- Pointing the zone apex at a load balancereasy
- Which identity is the AWS CLI using?easy
- What an Azure resource group iseasy
- Google Cloud resource hierarchyeasy
- Why use a managed databaseeasy
- Keeping CPU near a targeteasy
- How EBS snapshots store dataeasy
- Showing cloud cost per teameasy
- Shell access without opening port 22easy
- Well-Architected Framework pillarseasy
- Compute Savings Plans flexibilityeasy
- Which service model is App Service?easy
- Object storage on each cloudeasy
- Virtual machines on each cloudeasy
- Looking after the root accounteasy
- The least privilege principleeasy
- Picking an instance familyeasy
- What an AMI iseasy
- Auto Scaling group capacityeasy
- Data on an instance storeeasy
- Who deleted the bucket?easy
- Warning before the bill is higheasy
- What on-demand pricing meanseasy
- Keeping data in the UAEeasy
- What elasticity meanseasy
- Durability vs availabilityeasy
- Longest a Lambda can runeasy
- Lambda memory and CPUeasy
- What API Gateway doeseasy
- SNS vs SQSeasy
- Containers vs virtual machineseasy
- What Fargate removeseasy
- Managed Kubernetes on each cloudeasy
- Where container images liveeasy
- What kind of database DynamoDB iseasy
- What Cloud Run doeseasy
- Running code when a file arriveseasy
- Where to keep database passwordseasy
- What a key management service doeseasy
- Azure Functions billing on Consumptioneasy
- Coordinating several functionseasy
- What rightsizing meanseasy
- Disks left behindeasy
- Why teams move to Arm instanceseasy
- Storage class for unpredictable accesseasy
- A load balancer with no trafficeasy
- Snapshots that never expireeasy
- What a reserved instance iseasy
- Finding what drives the billeasy
- Google Cloud VPC scopeeasy
Medium 62
- When to use Spot instancesmedium
- Reducing serverless cold startsmedium
- Cutting NAT costs for S3 trafficmedium
- Reading an S3 SecureTransport policymedium
- SQS messages processed twicemedium
- Lambda in a VPC cannot reach an APImedium
- Automatic failover for RDSmedium
- Versioned bucket keeps growingmedium
- Azure VM reading Key Vault safelymedium
- Giving an app on EC2 access to S3medium
- An allow and a deny on the same actionmedium
- Reporting queries slow the main databasemedium
- Moving old logs to cheaper storagemedium
- Restoring from deep archivemedium
- Stopping accidental public bucketsmedium
- Instance up, app down, never replacedmedium
- No memory graph for a VMmedium
- Why define infrastructure as codemedium
- Blue-green deploymentsmedium
- Azure availability sets vs zonesmedium
- An automatic discount on Google Cloudmedium
- Reading a blob in the Archive tiermedium
- Which data transfer you pay formedium
- RPO vs RTOmedium
- Choosing a DR strategymedium
- Blocking resources outside allowed regionsmedium
- What a service control policy doesmedium
- General purpose or provisioned IOPS diskmedium
- Patching servers by replacing themmedium
- How much Lambda concurrency you needmedium
- Failed async Lambda eventsmedium
- Keeping order per customermedium
- One event, three consumersmedium
- Routing events by contentmedium
- Choosing a DynamoDB partition keymedium
- Query vs Scan in DynamoDBmedium
- Capacity mode for unpredictable trafficmedium
- Looking up orders by statusmedium
- How Aurora differs from standard RDSmedium
- Cosmos DB consistency levelsmedium
- Two roles on an ECS taskmedium
- Requests per Cloud Run instancemedium
- Uploading a 20 GB filemedium
- How KMS encrypts large datamedium
- Rotating a database password safelymedium
- A role nobody can assumemedium
- Where an Azure role assignment appliesmedium
- Downloaded service account keysmedium
- Expiring old session items automaticallymedium
- Turning dev servers off at nightmedium
- Monthly cost of a VM fleetmedium
- Monthly cost of a Lambda functionmedium
- Paying for data sent to usersmedium
- Moving volumes from gp2 to gp3medium
- Bringing existing licences to Azuremedium
- Log storage growing forevermedium
- Spotting a cost spike earlymedium
- Is a growing bill a problem?medium
- Using a CDN to cut origin costsmedium
- Azure NSG rule ordermedium
- Paying for public IPv4 addressesmedium
- Central network, many team projectsmedium
Hard 51
- S3 consistency modelhard
- Unexpected data transfer billhard
- Active-active writes across regionshard
- aws:SourceIp after adding a VPC endpointhard
- Cannot read objects a partner uploadedhard
- One NAT gateway for three AZshard
- DynamoDB throttling under the limithard
- Lambda spike breaks the databasehard
- Disk latency spike during nightly batchhard
- Replication skipped existing objectshard
- App writes to old primary after failoverhard
- How much Savings Plan to buyhard
- Moving a disk to another zonehard
- Scale-in stops jobs half-wayhard
- Cross-account bucket access still deniedhard
- A region missing a service you needhard
- Should a small team go multi-cloud?hard
- Migration bill higher than the data centrehard
- Scaling stopped at a hidden limithard
- Deleting an encryption keyhard
- Access keys pushed to a public repohard
- A brand-new role fails at firsthard
- Losing one of two zoneshard
- New instances take 12 minutes to servehard
- Uploads vanish after scalinghard
- A console fix that gets undonehard
- Who is responsible on a managed database?hard
- A function that triggers itselfhard
- Storing big documents in DynamoDBhard
- Two users book the same seathard
- One message blocks the consumershard
- Standard or Express workflowhard
- A report that takes 2 minuteshard
- Estimating Cosmos DB throughputhard
- Slow pods at 30% CPUhard
- Fargate tasks cannot pull imageshard
- A new item missing from the indexhard
- Events arrive out of orderhard
- An ML library too big for a ziphard
- Deploying from GitHub without stored keyshard
- Backups you cannot decrypt in the DR regionhard
- Serverless for steady heavy traffichard
- More memory, lower billhard
- Rightsized and now crashinghard
- Cheaper storage class, higher billhard
- The cost of a second regionhard
- Reservations stranded after a migrationhard
- Who should own cloud cost?hard
- Test environments nobody deleteshard
- Who secures a managed databasehard
- Private endpoint created, traffic still publichard