Easy ReviewDOP-C02
Resilient Cloud Solutions13

Elastic Disaster Recovery

Replicate server disks to AWS, rehearse recovery, and plan traffic failover and failback.

DRS keeps server disk changes in an AWS staging area and launches EC2 recovery instances when needed. Recovery plans also cover application validation, traffic routing, and returning data to the primary site.

Elastic Disaster Recovery basics

AWS Elastic Disaster Recovery replicates servers into AWS so they can be recovered quickly.

How it works

  1. Replicate source servers to AWS continuously.
  2. Keep low-cost storage and minimal compute while idle.
  3. Launch recovery instances when needed.

Exam cues

  • If the question says recover on-premises or cloud servers into AWS, think Elastic Disaster Recovery.

Remember

  • Low-cost standby, point-in-time recovery.

Trap: Elastic Disaster Recovery is for recovery; migration projects use AWS Application Migration Service.

From the AWS docs (5)
  • AWS Elastic Disaster Recovery reduces downtime and data loss with fast recovery of on-premises and cloud applications, using low-cost storage, minimal compute and point-in-time recovery.
  • Elastic Disaster Recovery recovers on-premises and cloud-based applications running on supported server operating systems.
  • The AWS Replication Agent continuously copies disk block changes from source servers to a target staging area.
  • The staging area subnet is in a selected AWS account and Region and uses storage and minimal compute for ongoing replication.
  • Recovery orchestration creates volumes from replicated data and converts non-AWS servers to boot natively on EC2.

Replication and staging

AWS Elastic Disaster Recovery (DRS) keeps copies of supported servers’ disk data in AWS so they can be recovered onto EC2. Its staging area stores replicated data with minimal compute while the source application continues running.

How it works

  1. Install the AWS Replication Agent on each supported source server. It copies disk block changes, meaning changes to portions of a disk, continuously toward AWS.
  2. Replication servers in the staging area subnetA staging area subnet contains the replication resources that maintain source-server data in AWS. write those changes to EBS volumes, the disks attached to EC2 instances. Allow TCP 1500 for replication data and TCP 443 for service communication.
  3. Maintain point-in-time recovery points, crash-consistentA crash-consistent recovery point contains disk data as if the server had stopped without flushing memory. snapshots that preserve data written to disk. Data still held only in memory is absent.
  4. For a drill or recovery, DRS creates recovery volumes and converts non-AWS servers to boot on EC2. The source server’s launch templateA launch template supplies EC2 settings such as the subnet for drill and recovery instances. selects the recovery subnet.

Exam cues

  • If the question says replicate supported on-premises servers to AWS with minimal standing compute, think DRS staging, because it maintains replicated storage before launching recovery instances.
  • If the question says replication falls behind after write traffic rises, think bandwidth and staging throughput, because replication must keep up with source disk writes.
  • If the question says recover an earlier server state, think point-in-time snapshotsA point-in-time snapshot preserves replicated disk data from a particular moment for recovery., because recovery can use an earlier disk recovery point.

Remember

  • RPORPO expresses how recent the recoverable replicated data is relative to the source., the recovery point objective, measures how much recent data may be missing. Seconds-level recovery points depend on replication capacity.
  • RTORTO expresses the time required to recover service after a disruption., the recovery time objective, concerns recovery time. DRS can launch in minutes; OS boot, applications, and target performance affect timing.
  • The fixed schedules retain points every 10 minutes for the prior hour and hourly for the prior 24 hours.
  • Daily snapshot retention can be set from 1 to 365 days.

Trap: Successful disk replication does not preserve memory-only application state. A crash-consistent recovery point contains data written to the replicated volumes.

Source disk changes reach staging storage
12Source serverReplicationStaging volumes
  1. The agent transfers disk changes to replication servers in the staging subnet.

  2. Replication servers write the received blocks to attached EBS volumes.

From the AWS docs (15)
  • The AWS Replication Agent continuously copies disk block changes from source servers to a target staging area.
  • The staging area subnet is in a selected AWS account and Region and uses storage and minimal compute for ongoing replication.
  • Replication servers write received blocks to attached EBS volumes.
  • Recovery orchestration creates volumes from replicated data and converts non-AWS servers to boot natively on EC2.
  • RPO measures how recent the last crash-consistent replicated data is; seconds-level RPO depends on sufficient network and staging throughput.
  • Application data held only in memory is not replicated until written to source volumes.
  • RTO is the time to recover service; DRS recovery timing depends on OS boot, applications, instance performance, and volume performance.
  • DRS can launch recovery instances in minutes using recent server state or a previous point in time.
  • DRS uses TCP 1500 from source servers to replication servers for data transfer and TCP 443 for service communication.
  • DRS point-in-time recovery uses crash-consistent EBS snapshots of replicated source volumes.
  • The fixed minute snapshot policy retains a recovery point every 10 minutes for the prior hour.
  • The fixed hourly snapshot policy retains hourly recovery points for the prior 24 hours.
  • Daily point-in-time snapshot retention is configurable from 1 to 365 days.
  • Drill and recovery instances use the subnet specified in the EC2 launch template associated with their source server.
  • Persistent replication lag can indicate insufficient network bandwidth and needs investigation.

Drills, failover, and failback

A DRS drill rehearses recovery while the source continues operating. FailoverFailover redirects traffic from a primary system to a recovery system. moves production traffic to recovery systems; failbackFailback redirects traffic from recovery systems to restored primary systems. returns it to the primary systems after their data catches up. Traffic redirection is a separate network operation.

How it works

  1. Check recovery readiness and healthy replication. Launch a drill and test the complete application, including dependency order and failback. Terminate drill instances when finished to stop their EC2 charges.
  2. During recovery, select recent data or an earlier point-in-time snapshotA point-in-time snapshot preserves replicated disk data from a particular moment for recovery. and launch recovery instances. Validate the applications and their dependencies.
  3. Redirect production traffic using your routing service. DRS does not perform that routing change.
  4. To return to the primary site, replicate changes made on the recovery systems back to the primary systems. For on-premises failback, boot the Failback Client on the destination server.
  5. Complete failback, validate the restored application, redirect traffic back, and confirm replication from the primary servers to AWS has resumed.

Exam cues

  • If the question says rehearse recovery without interrupting production or replication, think a DRS drill, because drill launches leave the source running.
  • If the question says recovery instances are healthy but users still reach the old site, think traffic redirection, because DRS launches instances without switching production routes.
  • If the question says return to the original site after accepting writes in AWS, think reverse replicationReverse replication copies recovery-system changes back to the primary systems before failback. first, because the primary systems need those new writes before serving users.

Remember

  • Recovery readiness checks include Ready, Healthy replication, and a successful prior drill.
  • The on-premises Failback Client supports x86_64. arm64 failback is supported within AWS.
  • Failback destination volumes must be at least as large as recovery volumes.
  • Disconnect from AWS removes replication resources and recovery points.

Trap: Booting a recovered server is only part of recovery. The application, its dependencies, and traffic routing must also work before users are switched over.

Return recovery-site writes before switching traffic
12Recovery instancePrimary serverTraffic routing
  1. Replicate recovery-site changes back and validate the restored primary system.

  2. Redirect users to the primary system outside DRS.

From the AWS docs (14)
  • DRS recovery drills do not disrupt source servers or their ongoing replication.
  • AWS recommends testing complete applications, recording launch order and dependencies, and including failback in drills.
  • Running drill instances incur EC2 charges and should be terminated after the drill.
  • Before recovery, DRS guidance checks Ready for recovery, Healthy replication, and evidence of a successful drill launch.
  • A recovery launch can use Use most recent data or a selected earlier point-in-time snapshot.
  • DRS launches recovery instances, but production traffic redirection is performed outside DRS.
  • AWS guidance validates recovered applications and dependencies before redirecting production traffic.
  • Failback redirects traffic from recovery systems to primary systems and is performed outside DRS.
  • Data written during recovery operation must be replicated back before redirecting users to the primary systems.
  • For on-premises failback, boot the Failback Client on the destination server to receive data from its recovery instance.
  • The Failback Client supports x86_64 servers; arm64 failback is supported within AWS, not to on-premises or another cloud.
  • On-premises failback destination volumes must be equal to or larger than the recovery instance volumes.
  • After failback, confirm the primary servers are replicating to AWS again.
  • Disconnect from AWS removes source replication resources, including point-in-time recovery points.

Flashcards

tap to flip
Answer

·

All 13 cards as a list

A company wants to recover on-premises servers into AWS quickly after a disaster, paying little while idle. Which service fits?

AWS Elastic Disaster Recovery.

When does AWS Elastic Disaster Recovery fit a recovery requirement?

When supported servers need continuous replication to AWS and recovery onto EC2. The staging area maintains replicated data with minimal compute.

On-premises servers need seconds-level data recovery points with limited compute running at the recovery site. Which AWS service fits?

Elastic Disaster Recovery. It continuously replicates disk blocks to a staging area and can launch recovery instances in minutes. Network and staging throughput must keep up with writes.

How does AWS Elastic Disaster Recovery turn replicated server disks into running recovery instances?

The agent copies disk changes to staging EBS volumes. Recovery orchestration creates recovery volumes, converts non-AWS servers, and boots EC2 instances.

A source server protected by AWS DRS writes faster than its available replication bandwidth. What happens to its recovery point?

It becomes older as replication falls behind. Seconds-level RPO requires enough network and staging throughput to keep up with source writes.

Does AWS Elastic Disaster Recovery preserve database updates that exist only in RAM when a server fails?

No. DRS replicates disk blocks. Data must reach the source volumes before it can be replicated.

A company needs to recover a server with AWS Elastic Disaster Recovery using an earlier state after data corruption. What recovery option fits?

Use point-in-time recovery. Elastic Disaster Recovery can launch a recovery instance from an earlier point using crash-consistent snapshots.

What do an AWS Elastic Disaster Recovery drill, failover, and failback each do?

A drill tests recovery without disrupting source replication. Failover redirects traffic to recovery systems. Failback redirects traffic to restored primary systems after reverse replication. Traffic routing happens outside DRS.

AWS DRS has launched recovery instances successfully, but users still reach the failed site. What step remains?

Redirect production traffic outside DRS, after application validation. Launching recovery instances does not change traffic routing.

A team uses AWS DRS recovery instances for several days, then plans to redirect users to its repaired primary site. What must happen first?

Replicate new recovery-site data back to the primary systems and validate the restored application before traffic redirection.

An AWS DRS drill boots each server successfully. Is this sufficient evidence that the application can recover?

No. Test the application as a whole, including dependencies and launch order, and rehearse failback.

How does AWS Elastic Disaster Recovery begin failback to an on-premises x86_64 server?

Boot the Failback Client on the destination server so it receives recovery-instance data. Destination volumes must be large enough.

After an AWS DRS failover, should operators immediately use Disconnect from AWS on the original source server?

No. Disconnect removes replication resources and point-in-time recovery points that may still be needed.

You finished every topic of DOP-C02!

Your badge turned gold. Next step: book the exam on AWS. Once you pass, add yourself to Who passed.

Share

Share that you finished this certification

I finished every topic of AWS Certified DevOps Engineer - Professional (DOP-C02) on Easy Review. Come study too!