Amazon EBS supplies disks for EC2 instances. Choose a volume type for the workload’s storage access pattern, then prepare snapshots and encryption keys so the data can be recovered in the required Availability Zone or Region.
EBS basics
Amazon Elastic Block Store (Amazon EBS) gives EC2 instances network-attached disks (block storageStorage presented to an instance as a disk for storing files or installing applications.) that keep data when the instance stops.
How it works
- Create a volume and attach it to an EC2 instance.
- The instance uses it like a local disk.
- SnapshotsA point-in-time backup that can restore the captured volume data to a new disk. back the volume up.
Exam cues
- If the question says a disk for one EC2 instance, think EBS.
Remember
- Block storage for EC2.
Trap: EBS is a disk for an instance; shared file storage across many instances is Amazon EFS.
Da documentação da AWS (11)
- Amazon EBS provides scalable, high-performance block storage volumes for use with Amazon EC2 instances.
- An instance store gives an EC2 instance temporary block storage on disks physically attached to the host, suited to buffers, caches and scratch data.
- An EBS snapshot is a point-in-time backup; each snapshot contains the information needed to restore the captured data to a new volume.
- EBS snapshots store only blocks changed since the preceding snapshot.
- Deleting an older EBS snapshot preserves blocks referenced by other snapshots.
- EBS snapshot charges are based on the data stored; because snapshots are incremental, only changed blocks add storage.
- EBS snapshots are stored in Amazon S3, in buckets the customer can't access directly.
- Neither st1 nor sc1 can be an EC2 boot volume.
- General Purpose SSD volumes are suitable for boot volumes.
- An EBS volume and the instance it attaches to must be in the same Availability Zone; EBS volumes persist independently of the instance.
- EBS charges for provisioned storage per GB-month by volume type, for provisioned IOPS and throughput above what is included, and for snapshot storage.
Volume types
Amazon EBS provides block storageStorage presented to an instance as a disk for storing files or installing applications., disks attached to EC2 instances. Choose solid-state driveA storage device used by EBS volume types suited to transactional reads and writes. (SSD) storage for transactional I/O, meaning application reads and writes, and hard disk driveA storage device used by EBS volume types suited to large sequential transfers. (HDD) storage for large sequential transfers.
How it works
- Identify the access pattern. IOPSThe number of storage read and write operations performed per second. means input/output operations per second. ThroughputThe amount of data a storage volume transfers per unit of time. is the amount of data transferred per second; MiB/s means mebibytes per second. Large transfers can reach a throughput limit before an IOPS limit.
- Choose a volume type from the table. Provision the size and performance the workload needs.
- Check the EC2 instance’s storage performance limits too. A faster volume cannot remove an instance bottleneck.
- For gp2 performance changes, consider Elastic Volumes to migrate to gp3 without interrupting the instance.
| Type | Performance model | Typical use |
|---|---|---|
| gp3 SSD | Performance independent of size; no burst creditsAn allowance that lets an eligible volume exceed its baseline performance temporarily. | General-purpose disks and boot volumes |
| gp2 SSD | Size-based baseline; smaller volumes can spend burst credits | Existing general-purpose disks |
| io2 Block Express SSD | Provisioned performance, low latencyThe elapsed time between sending a storage operation and receiving its completion acknowledgement. and higher durability | Demanding databases |
| io1 SSD | Provisioned IOPS | IOPS-intensive storage |
| st1 HDD | Throughput for frequent sequential I/O | Log processing and large sequential workloads |
| sc1 HDD | Lower-cost throughput for infrequent sequential I/O | Cold data |
Burst credits are an allowance for performance above a volume’s baseline. gp2 uses them; gp3 does not. Since April 2025, all io2 volumes use Block Express.
For Kubernetes on EKS, the standard EBS Container Storage InterfaceThe interface a storage driver uses to manage volumes for Kubernetes applications. (CSI) driver manages volume creation and attachment for application storage. Its controller needs an IAM roleAn AWS permissions identity a workload can use to make authorized API calls., an AWS permissions identity, associated through EKS Pod Identity or IAM roles for service accounts. EBS volumes cannot be mounted to Fargate Pods.
EKS Auto Mode manages block storage without installing that CSI controller. Its StorageClassA Kubernetes object defining how requested storage should be provisioned., a set of rules for provisioning disks, uses the separate ebs.csi.eks.amazonaws.com provisioner to set volume type and encryption.
Exam cues
- If the question says a small gp2 disk slows after sustained I/O, think BurstBalance, because its above-baseline performance consumes credits.
- If the question says storage size is sufficient but performance must increase, think gp3, because IOPS and throughput can be provisioned independently of size.
- If the question says large sequential logs are read frequently, think st1, because it targets frequent sequential transfers.
- If the question says sequential data is rarely read and cost matters, think sc1, because it targets infrequent access.
- If the question says EKS EBS provisioning fails with UnauthorizedOperation, think CSI controller permissions, because the driver calls AWS APIs.
Remember
- gp3 includes 3000 IOPS and 125 MiB/s.
- gp2 baseline is 3 IOPS per GiB, bounded by 100 and 16000 IOPS. Eligible small volumes can burst to 3000 IOPS.
- st1 and sc1 cannot boot an EC2 instance.
- For direct volume creation, the console defaults to gp3; command line tools, API and SDK default to gp2.
- EBS volume and instance: same AZAn isolated location within an AWS Region..
- st1: 0.125 to 16 TiB, baseline 40 MiB/s per TiB, burst bucket, 500 MiB/s maximum.
Trap: An EBS IOPS setting alone does not establish application performance. The I/O size, instance limits and workload demand also matter.
Da documentação da AWS (27)
- An EBS volume attaches to an EC2 instance and can be used like a local disk; EBS offers SSD and HDD volume types.
- IOPS measures input/output operations per second; I/O size also affects the throughput reached by a volume.
- gp3 lets you provision performance independently of volume size and does not use burst performance.
- gp3 includes a baseline of 3000 IOPS and 125 MiB/s of throughput.
- gp2 baseline performance scales at 3 IOPS per GiB, with a minimum of 100 and maximum of 16000 IOPS.
- Small gp2 volumes with a baseline below 3000 IOPS can burst to 3000 IOPS while I/O credits last; BurstBalance reports remaining credits.
- io2 Block Express targets demanding workloads with low latency and higher durability than General Purpose SSD.
- Since April 30, 2025, all new and existing io2 volumes use Block Express.
- Throughput Optimized HDD (st1) targets frequently accessed, large sequential workloads such as log processing.
- Cold HDD (sc1) targets infrequently accessed large sequential data at lower cost than st1.
- Realized EBS performance depends on instance configuration and I/O demand, not only the selected volume performance.
- Elastic Volumes can migrate a gp2 volume to gp3 and modify type, IOPS and throughput without interrupting its EC2 instance.
- The create-volume console defaults to gp3; command line tools, API and SDK default to gp2.
- The EBS Container Storage Interface (CSI) driver manages EBS volume lifecycles for Kubernetes storage and needs IAM permissions to call AWS APIs.
- Configure EBS CSI driver permissions with EKS Pod Identity or IAM roles for service accounts, associating the role with the controller identity.
- Amazon EBS volumes cannot be mounted to EKS Fargate Pods.
- io1 volumes provide a configured IOPS rate for I/O-intensive workloads, particularly databases sensitive to consistent storage performance.
- EKS Auto Mode does not require installing the EBS CSI controller and uses the ebs.csi.eks.amazonaws.com provisioner separately from the standard CSI driver.
- An EKS Auto Mode StorageClass defines volume provisioning settings such as volume type and encryption.
- The DeleteOnTermination attribute decides whether an EBS volume is deleted when its instance terminates: a root volume attached at launch is deleted by default, while a data volume attached after launch is preserved.
- io1 volumes allow at most 50 provisioned IOPS per GiB, from 100 to 64,000 IOPS per volume; io2 Block Express allows 1,000 IOPS per GiB, up to 256,000 IOPS.
- gp3 can be provisioned up to 80,000 IOPS (500 IOPS per GiB) and 2,000 MiB/s of throughput.
- gp2 volumes range from 1 GiB to 16 TiB and gp3 volumes from 1 GiB to 64 TiB.
- io1 volumes range from 4 GiB to 16 TiB; io2 Block Express goes up to 64 TiB and 4,000 MiB/s of throughput.
- RAID 0 stripes I/O across EBS volumes so their IOPS and throughput add up; AWS does not recommend RAID 5 or 6 on EBS because parity writes cost 20 to 30 percent of usable IOPS.
- EBS replicates volume data across multiple servers in its Availability Zone; io2 Block Express is designed for 99.999 percent durability and other types for 99.8 to 99.9 percent.
- st1 volumes range from 0.125 TiB to 16 TiB and use a throughput burst bucket: baseline 40 MiB/s per TiB, with throughput capped at 500 MiB/s per volume.
Multi-Attach
EBS Multi-Attach connects one io1 or io2 volume to several EC2 instances in the same Availability ZoneAn isolated location within an AWS Region. (AZ). It supports applications designed to coordinate access to shared block storageStorage presented to an instance as a disk for storing files or installing applications.; it does not provide that coordination for them.
How it works
- Enable Multi-Attach on an eligible io1 or io2 volume and attach it to supported Nitro-based instances, EC2 instances built on AWS’s Nitro platform, in the same AZ.
- Each attached instance can read and write the volume. All attachments share its provisioned performance.
- Use a clustered file systemA file system designed to coordinate access to a shared disk from multiple servers., software designed to coordinate shared disk access, and application write ordering. I/O fencingA mechanism that controls write access to shared storage to preserve data consistency. controls which instances can write so a conflicting writer cannot damage shared data.
- Set consistent delete-on-termination behavior. The last attached instance’s setting determines deletion when that instance terminates.
| Requirement | Support |
|---|---|
| Volume types | io1 and io2 |
| Linux | io1 and io2 |
| Windows | io2 only |
| Placement | Same AZ; up to 16 Nitro-based instances |
| Boot volumeThe disk containing the data an EC2 instance uses to boot. | Unsupported |
Exam cues
- If the question says an application coordinates shared disk writes within one AZ, think Multi-Attach, because supported instances can access one io1 or io2 volume.
- If the question says ordinary EXT4 or XFS should be mounted by concurrent writers, think a clustered file system, because standard file systems do not coordinate this access.
- If the question says instances in different AZs must attach to one volume, think an unmet placement requirement, because Multi-Attach stays within one AZ.
Remember
- Multi-Attach supports up to 16 Nitro-based instances per volume in the same AZ.
- Windows supports io2 only; Linux supports io1 and io2.
- An EBS infrastructure issue can affect every attachment to the shared volume.
- Detach the volume before enabling or disabling Multi-Attach.
Trap: Multi-Attach adds attachments, not independent copies. It does not turn one AZ’s volume into storage shared across AZs.
Da documentação da AWS (10)
- Multi-Attach lets multiple EC2 instances read and write the same io1 or io2 volume in one Availability Zone.
- A Multi-Attach volume can attach to up to 16 Nitro-based instances in the same AZ.
- Linux supports Multi-Attach io1 and io2; Windows supports Multi-Attach io2 only.
- Standard EXT4 and XFS file systems are not designed for simultaneous access from multiple servers; use a clustered file system.
- Applications must order writes across attached instances; I/O fencing controls write access to preserve data consistency.
- Multi-Attach volumes cannot be boot volumes.
- An issue at the EBS infrastructure layer can make a Multi-Attach volume unavailable to every attached instance.
- Provisioned Multi-Attach volume performance is shared across all attached instances.
- The last attached instance termination determines volume deletion according to its delete-on-termination setting.
- Multi-Attach cannot be enabled or disabled while the volume is attached to an instance.
Snapshots and cross-AZ recovery
An EBS snapshotA point-in-time backup that can restore the captured volume data to a new disk. is a point-in-time backup used to create a replacement disk. The volume must share its instance’s Availability ZoneAn isolated location within an AWS Region. (AZ), but the snapshot can restore volumes across AZs in the same RegionA separate geographic area containing multiple isolated Availability Zones..
How it works
- Prepare the application for the snapshot. Pause writes or unmountMake a disk file system unavailable to the operating system before detaching or preparing it for backup. the volume, making its file system unavailable to the operating system. AWS recommends stopping the instance for a root-volume snapshot, which backs up its boot disk.
- Request the snapshot. It captures data already written to disk, not data still in application or operating system caches. You can resume use while its data transfer is pending.
- EBS stores incremental snapshot data: only blocks changed since the preceding snapshot. Each snapshot still has the information needed to restore its captured state.
- In the recovery AZ, create a new volume from the snapshot and attach it to an instance in that AZ. MountMake a disk file system accessible to the operating system. it, making its file system accessible to the operating system.
- Account for volume initializationLoading snapshot data blocks onto a restored volume so it can deliver full performance., the loading of snapshot blocks onto the new volume, before expecting full performance. For regional recovery, copy the completed snapshot to the destination Region first.
For an attached disk that cannot complete I/O, inspect the EC2 attached-EBS status check. Its metric is StatusCheckFailed_AttachedEBS. For an underlying EC2 system problem, stopping and starting an EBS-backed instance usually moves it to another host; this does not move its volume to another AZ.
Exam cues
- If the question says recover an EBS disk in another AZ, think snapshot then a new volume, because a volume can attach only within its AZ.
- If the question says delete old snapshots without damaging recent recovery points, think incremental block references, because EBS preserves blocks used by remaining snapshots.
- If the question says application data was still cached at snapshot time, think missing unwritten data, because a snapshot captures the volume’s stored data.
Remember
- Volume attachment is scoped to an AZ; regional snapshot recovery can choose any AZ in that Region.
- Deleting a snapshot does not delete data referenced by other snapshots.
- EBS can snapshot multiple volumes attached to an instance.
- A restored volume may need initialization before it delivers full performance.
Trap: An incremental snapshot is still a complete recovery point. You do not manually combine snapshot files to restore a volume.
Create a snapshot of the source volume. Pause writes first when application consistency matters.
Create a volume from the snapshot in the target AZ and attach it to an instance there.
Detalhes das caixas (3)
- EBS volume
An EBS volume lives in one AZ. You cannot attach it to an instance in another AZ.
- EBS snapshot
Snapshots are incremental and stored regionally, so any AZ in the Region can restore them. Copy the snapshot to reach another Region.
- New volume
Restore in the target instance AZ. Full performance may require initialization. FSR must be enabled for this snapshot and target AZ.
Da documentação da AWS (12)
- EBS snapshot data is replicated across AZs in its Region and can restore a volume in any AZ there.
- An EBS volume must be created in the same AZ as the EC2 instance to which it attaches.
- EBS snapshots include data written to the volume when requested, excluding application and operating system caches.
- AWS recommends pausing writes or unmounting a volume before snapshot creation; for a root volume, stop the instance first.
- Snapshot creation is asynchronous; data transfers while the snapshot is pending, and volume use can resume after the request.
- Recovery from an EBS snapshot creates a replacement volume; attach and mount it on the target instance.
- Copy a completed EBS snapshot to another Region before using that copy for recovery there.
- Snapshot-restored volumes must download blocks before full performance; use FSR, a provisioned initialization rate, or manual initialization to manage this.
- FSR must be enabled on the specific snapshot in the AZ used to create the restored volume.
- Attached EBS status checks detect volume reachability and I/O impairment; StatusCheckFailed_AttachedEBS signals an attached volume cannot complete I/O.
- EBS supports taking multi-volume snapshots of volumes attached to an EC2 instance.
- Stopping and starting an EBS-backed EC2 instance usually migrates it to another host when addressing an underlying system problem.
Snapshot lifecycle management
Data Lifecycle Manager (DLM) automates EBS snapshotA point-in-time backup that can restore the captured volume data to a new disk. and EBS-backed Amazon Machine ImageAn image used to launch an EC2 instance with its saved storage and configuration. (AMI) management. An AMI is the machine image used to launch an EC2 instance. DLM applies a schedule and retentionA rule deciding how many backups to keep or how long to keep them. rules, which decide how long backups remain.
How it works
- Select source volumes or instances with tags, resource labels written as key-value pairs. An instance target creates a multi-volume snapshot set; a volume target backs up that volume.
- Choose a schedule and retention by count or age. The first run begins within an hour after its scheduled time in UTC, the reference time zone used by the schedule.
- For application-consistent snapshotsA snapshot taken after application I/O is prepared so buffered data is captured consistently., use an instance-targeted policy and Systems Manager pre and post scripts to pause application I/O, flush buffered data and resume normal I/O.
- Configure copies in another RegionA separate geographic area containing multiple isolated Availability Zones., sharing with other accounts or fast snapshot restore (FSR), which prepares snapshot restores for immediate volume performance.
| Need | Mechanism | Scope |
|---|---|---|
| Scheduled EBS snapshots and cleanup | DLM policy | DLM-created resources |
| One backup plan across supported services | AWS Backup | EBS and other supported services |
Exam cues
- If the question says tagged EBS volumes need scheduled snapshots, cleanup and regional copies, think DLM, because its policy covers that lifecycle.
- If the question says one backup plan spans storage and databases, think AWS Backup, because it supports multiple services.
Remember
- Configure backups; volume replication is not an automatic snapshot policy.
- DLM manages only resources it created, including EBS-backed AMIs.
- Retention can use count or age, with one retention type across a policy’s schedules.
- DLM has no additional service fee. Snapshot storage, copies and FSR can incur charges.
Trap: A DLM policy cannot clean up snapshots created manually or by AWS Backup. Manage those through the mechanism that owns them.
Da documentação da AWS (14)
- AWS does not automatically back up EBS data; create snapshots or configure snapshot automation.
- Data Lifecycle Manager (DLM) automates creation, retention and deletion of EBS snapshots and EBS-backed AMIs.
- A custom DLM snapshot policy selects source volumes or instances using matching tag key-value pairs.
- A DLM policy targeting an instance creates a multi-volume snapshot set; targeting a volume snapshots that volume.
- DLM snapshot retention can use snapshot count or age; all schedules in a policy share the retention type.
- DLM custom policies support daily, weekly, monthly, yearly or cron schedules; the first run begins within an hour after its scheduled UTC time.
- DLM snapshot policies can configure cross-Region copies, cross-account sharing and FSR in selected AZs.
- DLM cannot manage snapshots or AMIs created by other means, and does not manage instance-store-backed AMIs.
- DLM has no additional service charge; storage and other enabled features can still incur charges.
- AWS Backup centrally configures, schedules and monitors backups for EBS and other supported AWS services.
- DLM pre and post scripts can pause application I/O and flush buffered data before snapshots, then resume normal I/O after creation.
- DLM application-consistent pre and post scripts require snapshot policies targeting instances and run through Systems Manager command documents.
- Recycle Bin keeps deleted EBS volumes, EBS snapshots and EBS-backed AMIs for a retention period you set, so they can be restored before they are permanently deleted.
- EBS Snapshots Archive costs up to 75 percent less for snapshots kept 90 days or longer and rarely accessed; an archived snapshot must be restored to the standard tier before it can be used.
Fast snapshot restore
Fast snapshotA point-in-time backup that can restore the captured volume data to a new disk. restore (FSR) prepares an EBS snapshot so restored volumes can start fully initialized. Initialization means loading snapshot blocks onto the volume. FSR avoids the latencyThe elapsed time between sending a storage operation and receiving its completion acknowledgement. of loading a block on its first access, within its supported performance limits.
How it works
- Enable FSR for the specific recovery snapshot in each target Availability ZoneAn isolated location within an AWS Region. (AZ).
- Wait for its
enabledstate and check volume creation creditsThe per-snapshot and per-AZ allowance for creating volumes with FSR benefits., the allowance for creating initialized volumes from that snapshot in that AZ. - Create the volume in the enabled AZ. Each FSR restore consumes one credit; insufficient credits create the volume without FSR benefits.
- Prepare each new snapshot or copy separately. DLM can enable FSR for snapshots produced by a schedule.
| Initialization method | When data loads | Useful requirement |
|---|---|---|
| FSR | Prepared for fully initialized volume creation | Immediate restored performance within FSR limits |
| Provisioned initialization rate | After creation at 100 to 300 MiB/s | Predictable initialization time |
| Manual initialization | Read volume blocks after creation | Prepare the restored disk before workload use |
Exam cues
- If the question says a restored database is slow on initial reads and must start at full performance, think FSR, because the restored volume starts initialized within its supported limits.
- If the question says a copy of an FSR snapshot lacks the benefit, think explicit enablement, because copies do not inherit FSR.
Remember
- Scope: one snapshot and one AZ per FSR enablement.
- Quota: 5 enabled snapshots per RegionA separate geographic area containing multiple isolated Availability Zones., including snapshots shared with the account.
- Billing: per minute per snapshot-AZ pair, with a one-hour minimum.
- Snapshot size: at most 16 TiB.
- Full performance benefit: up to 64000 IOPSThe number of storage read and write operations performed per second. and 1000 MiB/s. Above these, initialize the volume as AWS recommends.
Trap: An enable request is not enough. Check
enabledstate, credits, target AZ and supported volume performance before relying on FSR for recovery.
Da documentação da AWS (13)
- A snapshot-restored EBS volume downloads snapshot blocks before full performance; initialization can increase I/O latency.
- Fast snapshot restore (FSR) creates a fully initialized volume from a snapshot, avoiding the first-block-access latency penalty within supported performance limits.
- Enable FSR separately for each snapshot and target AZ; restore the volume in an enabled AZ.
- Snapshot copies and new snapshots of restored volumes do not inherit FSR.
- FSR has a quota of 5 enabled snapshots per Region, including owned and shared snapshots enabled by the account.
- FSR is billed per minute for each enabled snapshot-AZ pair, with a minimum of one hour.
- FSR supports snapshots of 16 TiB or less.
- Full FSR benefit applies up to 64000 provisioned IOPS and 1000 MiB/s; above these levels, AWS recommends initializing the volume for full performance.
- Each FSR snapshot-AZ pair has volume creation credits; each restored volume consumes one credit and insufficient credits cause creation without FSR benefits.
- The FSR enabled state with sufficient volume creation credits provides the full restoration performance benefit.
- You can manually initialize a snapshot-restored volume by reading its blocks before using it for the workload.
- A provisioned initialization rate from 100 to 300 MiB/s downloads snapshot blocks after creation for predictable initialization time.
- DLM snapshot schedules can enable FSR in selected AZs and control how long or how many snapshots retain it.
Encryption and Regional copies
EBS encryption protects stored data and the transfer between an EC2 instance and its attached disk. It uses AWS KMS keys, encryption keys managed through AWS Key Management Service. Regional recovery adds a snapshotA point-in-time backup that can restore the captured volume data to a new disk. copy in the destination RegionA separate geographic area containing multiple isolated Availability Zones. and access to a suitable encryption key.
How it works
- Create a snapshot and wait for its
completedstate before copying it. Copy the snapshot to the recovery Region; the copy has its own resource ID. - Select an accessible destination-Region KMS key for the copy when required. You can encrypt an unencrypted snapshot copy or use a different key for an encrypted copy.
- Create the replacement volume in the recovery instance’s Availability ZoneAn isolated location within an AWS Region. (AZ). An encrypted snapshot produces an encrypted volume.
- For replacement on an existing instance, unmountMake a disk file system unavailable to the operating system before detaching or preparing it for backup. and detach the old disk, then attach and mountMake a disk file system accessible to the operating system. the replacement. You change the data’s encryption by creating a replacement resource, not by changing an existing disk’s key.
To encrypt an existing unencrypted disk, take a snapshot, make an encrypted copy and restore a replacement volume. You can also create an encrypted replacement directly from the unencrypted snapshot.
| Requirement | Mechanism | Constraint |
|---|---|---|
| Encrypt future volumes and snapshot copies | EBS encryption by default | Per account and Region; existing resources stay unchanged |
| Change an existing resource’s key | Snapshot copy with another KMS key | Creates a new resource |
| Share an encrypted snapshot | Customer managed keyAn encryption key a customer manages and can authorize other accounts to use. plus snapshot and key permissions | Default AWS managed key snapshots cannot be shared |
| Evaluate attached-volume encryption | AWS Config encrypted-volumes |
Optional kmsId checks the expected key |
Exam cues
- If the question says recover encrypted EBS data in another Region, think snapshot copy and destination key access, because the source snapshot is regional.
- If the question says enabling encryption by default should fix existing disks, think encrypted replacement volumes, because the setting only affects new resources.
- If the question says an encrypted snapshot must be shared with another account, think a customer managed key and both permissions, because snapshot permission alone is insufficient.
Remember
- You cannot change the key attached to an existing EBS volume or snapshot.
- Snapshot copies can use a different key.
- Encryption by default is Region-specific.
- EBS does not remove encryption when restoring or copying encrypted snapshots.
- AWS Config’s encryption rule evaluates attached volumes.
Trap: A shared encrypted snapshot still requires key access. A snapshot encrypted with the default AWS managed key must first be copied under a customer managed key before sharing.
Copy the completed snapshot into the recovery Region with an accessible destination key.
Create an encrypted volume in the recovery instance AZ and attach it there.
Da documentação da AWS (13)
- EBS encryption uses AWS KMS encryption keys and protects data at rest and data in transit between an EC2 instance and its attached volume.
- Snapshots are regional; copy a snapshot to the destination Region before sharing or restoring it there.
- Copy an EBS snapshot after it reaches completed state; the copy has its own resource ID.
- You cannot change the key on an existing EBS volume or snapshot; copying a snapshot can encrypt the copy with a different key.
- Create a snapshot of an unencrypted volume, copy the snapshot with encryption, then create an encrypted replacement volume from the copy.
- An encrypted replacement volume can also be created directly from an unencrypted snapshot by enabling encryption during volume creation.
- After creating a replacement EBS volume, unmount and detach the old volume, attach the replacement in the same AZ, and mount it.
- Encryption by default is an account setting per Region that applies to new volumes and snapshot copies; it does not change existing resources.
- A snapshot from an encrypted volume inherits encryption and its KMS key; data restored from an encrypted snapshot remains encrypted.
- EBS cannot remove encryption when copying an encrypted snapshot or restoring a volume from it.
- Encrypted EBS snapshot sharing requires a customer managed key and access to both the snapshot and the key; default AWS managed key snapshots cannot be shared.
- DLM cross-Region snapshot copies use a KMS key in the destination Region; a selected key must be accessible to the policy role.
- The AWS Config encrypted-volumes rule evaluates attached EBS volumes and can flag missing encryption or use of a key outside its optional kmsId parameter.
Flashcards
Nada para revisar aqui agora. Escolha outros tópicos ou outro filtro.
·
Todos os 49 flashcards em lista
An EC2 instance needs a persistent disk for its operating system and database. Which AWS storage service fits?
Amazon EBS, block storage volumes attached to EC2 instances.
An EC2 workload needs very fast scratch disk space whose data can be lost. EBS or instance store?
Instance store, temporary block storage on disks attached to the host. Data that must persist belongs on EBS.
Do incremental EBS snapshots require retaining every older snapshot to restore a full volume?
No. Each snapshot has the information needed to restore its captured data, and deleting an older snapshot preserves blocks used by other snapshots.
A 500 GiB EBS volume has 20 GiB of changes since its last snapshot. Does the new snapshot add 500 GiB of storage charges?
No. Snapshots are incremental and billed on data stored, so the new snapshot adds only the changed blocks.
Can an EC2 instance boot from an EBS st1 or sc1 HDD volume?
No. Neither HDD type supports boot volumes. General Purpose SSD supports boot use.
Can an EC2 instance in us-east-1a attach an EBS volume created in us-east-1b?
No. An EBS volume attaches only to instances in its own Availability Zone. Use a snapshot to create a volume in the other AZ.
Which factors determine the charge for an Amazon EBS volume?
Provisioned storage per GB-month by volume type, provisioned IOPS and throughput above what the volume type includes, and snapshot storage.
When should an EC2 workload choose EBS gp3 rather than io2 Block Express?
Use gp3 for general-purpose SSD storage with independently configured performance. Choose io2 Block Express for demanding workloads that need lower latency and higher durability.
A small EBS gp2 volume slows after a period of heavy database I/O. Which metric and storage behavior should the team inspect?
Inspect BurstBalance. gp2 can spend credits to exceed its size-based baseline; performance falls back when credits run out.
An EC2 workload needs more EBS IOPS without buying extra disk space or depending on burst credits. Which general-purpose volume type fits?
gp3. Its performance is configured independently of size and it does not use burst credits.
Which EBS HDD types fit frequent sequential log processing versus infrequently read sequential data?
st1 for frequently accessed sequential workloads; sc1 for infrequently accessed sequential data where lower cost matters.
What EBS gp3 performance is included before provisioning extra IOPS or throughput?
3000 IOPS and 125 MiB/s.
An EKS application cannot provision EBS storage and reports UnauthorizedOperation. What should the team check?
The EBS CSI controller needs an IAM role with EBS API permissions, connected through EKS Pod Identity or IAM roles for service accounts.
Can an EKS Fargate Pod mount an EBS volume through the EBS CSI driver?
No. EBS volumes cannot be mounted to Fargate Pods.
An EC2 workload needs more EBS performance without increasing volume size. Which volume type supports this?
Use gp3. Its performance can be provisioned independently of volume size.
A team must keep an instance root volume after the instance is terminated. What do they change?
Set DeleteOnTermination to false on the root volume; by default a root volume attached at launch is deleted with the instance.
What is the most IOPS a 10 GiB io1 volume can be provisioned with?
500 IOPS, because io1 allows at most 50 IOPS per GiB.
What are the maximum IOPS and throughput of a gp3 volume?
80,000 IOPS (at 500 IOPS per GiB, so from 160 GiB) and 2,000 MiB/s.
A database needs more IOPS than one EBS volume can give. What block-level approach adds performance across volumes?
A RAID 0 stripe across several EBS volumes, up to the instance EBS bandwidth limit. Avoid RAID 5 or 6, which waste IOPS on parity.
What is the maximum throughput of an Amazon EBS Throughput Optimized HDD (st1) volume, and what limits a small one?
500 MiB/s. A small st1 volume has a low baseline (40 MiB/s per TiB) and bursts on throughput credits.
Can several EC2 instances share any EBS volume with Multi-Attach?
No. Multi-Attach supports io1 or io2 with up to 16 Nitro-based instances in the same AZ. Linux supports both; Windows supports io2 only. Coordinate concurrent writes.
Two EC2 instances mount the same Multi-Attach EBS volume using ordinary EXT4 and write concurrently. Is attachment permission enough to preserve the data?
No. Use a clustered file system and coordinate write ordering. Ordinary EXT4 and XFS do not support this shared-writer design.
An application needs two EC2 instances in different AZs to share one EBS io2 Multi-Attach volume. Does Multi-Attach meet the requirement?
No. All attached instances must be in the same AZ as the volume.
An EBS Multi-Attach volume is provisioned for a given IOPS rate. Does every attached EC2 instance receive that entire rate independently?
No. The volume performance is shared across attached instances.
What determines whether an EBS Multi-Attach volume is deleted when its final attached instance terminates?
The delete-on-termination setting in the final instance attachment. Use consistent settings across attachments for predictable behavior.
Can an EBS volume attach to an EC2 instance in a different Availability Zone?
No. Create a snapshot, restore a new volume in the target instance AZ, then attach that new volume.
How should a team prepare an EBS snapshot when application consistency matters?
Pause writes or unmount the volume before requesting the snapshot. Data still in application or operating system caches is not included; stop the instance first for a root-volume snapshot.
An AZ fails and an EC2 workload must recover its EBS data in a different AZ in the same Region. Is a cross-Region snapshot copy required?
No. Restore a new volume from the regional snapshot in the target instance AZ.
An EBS snapshot was requested while recent database data remained in memory. Will the snapshot include those unwritten changes?
No. The snapshot includes data already written to the volume when requested, not application or operating system caches.
An EC2 instance needs data restored from an EBS snapshot in another Availability Zone in the same Region. What should the team do?
Create a volume from the snapshot in the target Availability Zone, then attach and mount it on the instance.
A disaster recovery plan needs an EBS snapshot available in another AWS Region. What must happen before recovery there?
Copy the completed snapshot to the recovery Region, then restore a volume from that copy.
Does AWS automatically create backups for an EBS volume with no configured backup policy?
No. Create snapshots yourself or configure DLM or AWS Backup.
What does EBS Data Lifecycle Manager automate?
The creation, retention and deletion of EBS snapshots and EBS-backed AMIs. Snapshot policies can select tagged resources, copy across Regions, share across accounts and enable FSR.
Can DLM apply its lifecycle policy to EBS snapshots created manually or by AWS Backup?
No. DLM cannot manage snapshots created by other means.
Tagged EBS volumes need nightly snapshots, a fixed number of retained snapshots and a copy in another Region. Which tool fits?
A DLM custom snapshot policy targeting the resource tags, with a nightly schedule, count-based retention and a cross-Region copy rule.
An AWS backup requirement covers EBS, EFS and databases under one centrally managed plan. Should the team use only DLM?
Use AWS Backup for supported services. DLM focuses on EBS snapshots and EBS-backed AMIs.
A database needs scheduled EBS snapshots with buffered writes flushed before capture. Which DLM feature fits?
Pre and post scripts for application-consistent snapshots. Prepare application I/O before the snapshot and resume it afterward.
An admin deleted critical EBS snapshots by mistake. Which feature, enabled beforehand, would have allowed restoring them?
Recycle Bin with a retention rule for EBS snapshots.
Compliance snapshots must be kept for years and are almost never restored. How can their storage cost drop?
Move them to the EBS Snapshots Archive tier, up to 75 percent cheaper for snapshots kept 90 days or more; restore one to the standard tier before using it.
What does EBS fast snapshot restore provide, and how is it scoped and billed?
A fully initialized restored volume within supported limits. Enable it per snapshot and AZ. Billing is per minute per snapshot-AZ pair with a one-hour minimum; the quota is 5 enabled snapshots per Region.
A team copies an FSR-enabled EBS snapshot and creates a new snapshot of its restored volume. Do the new snapshots inherit FSR?
No. Enable FSR explicitly for each snapshot in each required AZ.
An EBS-restored database must avoid the first-read performance penalty during failover. What should the team prepare?
Enable FSR for the recovery snapshot in the recovery AZ before creating the volume, check enabled state and creation credits, and stay within supported performance limits. DLM can automate FSR for new snapshots.
An EBS recovery plan allows time to initialize disks but needs a predictable download rate. Which alternative to FSR fits?
Specify a provisioned initialization rate of 100 to 300 MiB/s when creating volumes from snapshots. The download still happens after creation.
How can an EC2 workload recover an encrypted EBS volume in another AWS Region?
Copy its completed snapshot to the target Region using an accessible destination key, then create a new encrypted volume in the target instance AZ.
How can a team encrypt an existing unencrypted EBS volume using a snapshot copy?
Snapshot it, create an encrypted snapshot copy, restore a new encrypted volume, then replace the original volume. A direct encrypted restore from the unencrypted snapshot is also supported.
An account enables EBS encryption by default in one Region. Are its existing volumes encrypted and its other Regions covered automatically?
No. The setting is Region-specific and does not modify existing volumes or snapshots.
An encrypted EBS snapshot uses the default AWS managed key. Can its owner share it with a recovery account?
No. Copy it under a customer managed key, then share the snapshot and grant the recovery account access to that key.
Which AWS Config rule can identify attached EBS volumes that are unencrypted or use an unexpected KMS key?
encrypted-volumes. Its optional kmsId parameter specifies the expected encryption key.
Can an encrypted EBS snapshot be copied into an unencrypted snapshot to simplify regional recovery?
No. EBS preserves encryption for copies and restored volumes from encrypted snapshots.