Amazon EBS FAQs

When attached to EBS-optimized instances, Provisioned IOPS SSD (io2 Block Express and io1) volumes are designed to deliver within 10% of the provisioned IOPS performance 99.9% of the time in a given year. Your exact performance depends on your application’s I/O requirements.

When attached to EBS-optimized instances, Provisioned IOPS io2 Block Express volumes can achieve sub-millisecond latency, and io1 volumes can achieve single-digit millisecond latencies. Your exact performance depends on your application’s I/O requirements.

Yes, it does. When you provision IOPS for io2 Block Express volumes, the IOPS rate you get depends on the I/O size of your application reads and writes. Provisioned IOPS volumes have a base I/O size of 16KB. So, if you have provisioned a volume with 40,000 IOPS for an I/O size of 16KB, it will achieve up to 40,000 IOPS at that size. If the I/O size is increased to 256 KB, then you will achieve up to 16,000 IOPS, since the maximum throughput of 4000 MiB/s is achieved at 16,000 IOPS. For more details, please visit the technical documentation on Provisioned IOPS volumes. You can use Amazon CloudWatch to monitor your throughput and I/O sizes.

Provisioned IOPS SSD (io2 Block Express and io1) volumes attached to EBS-optimized instances are designed to offer consistent performance, delivering within 10% of the provisioned IOPS performance 99.9% of the time over a given year. For maximum performance consistency with new volumes created from a snapshot, we recommend enabling Fast Snapshot Restore (FSR) on your snapshots. EBS volumes restored from FSR-enabled snapshots instantly receive their full performance.

Another factor that can impact your performance is if your application isn’t sending enough I/O requests. This can be monitored by looking at your volume’s queue depth. The queue depth is the number of pending I/O requests from your application to your volume. For maximum consistency, a Provisioned IOPS volume must maintain an average queue depth (rounded to the nearest whole number) of one for every 1000 provisioned IOPS in a minute. For example, for a volume provisioned with 3000 IOPS, the queue depth average must be 3. For more information about ensuring consistent performance of your volumes, see Increasing EBS Performance.

When attached to EBS-optimized instances, Throughput Optimized HDD (st1) and Cold HDD (sc1) volumes are designed to deliver within 10% of the expected throughput performance 99% of the time in a given year. Your exact performance depends on your application’s I/O requirements and the performance of your EC2 instance.

Yes. The throughput rate you get depends on the I/O size of your application reads and writes. HDD-backed volumes process reads and writes in I/O sizes of 1MB. Sequential I/Os are merged and processed as 1 MB units while each non-sequential I/O is processed as 1MB even if the actual I/O size is smaller. Thus, while a transactional workload with small, random IOs, such as a database, won't perform well on HDD-backed volumes, sequential I/Os and large I/O sizes will achieve the advertised performance of st1 and sc1 for a longer period of time.

Throughput Optimized HDD (st1) and Cold HDD (sc1) volumes attached to EBS-optimized instances are designed to offer consistent performance, delivering within 10% of the expected throughput performance 99% of the time in a given year. There are several factors that could affect the level of consistency you see. For example, the relative balance between random and sequential I/O operations on the volume can impact your performance. Too many random small I/O operations will quickly deplete your I/O credits and lower your performance down to the baseline rate. Your throughput rate may also be lower depending on the instance selected. Although st1 can drive throughput up to 500 MB/s, performance will be limited by the separate instance-level limit for EBS traffic. Another factor is taking a snapshot which will decrease expected write performance down to the baseline rate, until the snapshot completes. This is specific to st1 and sc1.

Your performance can also be impacted if your application isn’t sending enough I/O requests. This can be monitored by looking at your volume’s queue depth and I/O size. The queue depth is the number of pending I/O requests from your application to your volume. For maximum consistency, HDD-backed volumes must maintain an average queue depth (rounded to the nearest whole number) of four or more for every 1 MB sequential I/O. For more information about ensuring consistent performance of your volumes, see Increasing EBS Performance.

Yes. You can stripe multiple volumes together to achieve up to 400,000 IOPS or 12,500 Mbps when attached to larger EC2 instances. We recommend using io2 Block Express volumes for higher performance requirements without needing the operational management of striping multiple volumes. Performance for st1 and sc1 scales linearly with volume size so there may not be as much of a benefit to stripe these volumes together.

EBS is a multi-tenant block storage service. We employ rate limiting as a mechanism to avoid resource contention. This starts with having defined performance criteria for the volumes – our volume types (gp2, PIOPS, st1, and sc1) all have defined performance characteristics in terms of IOPS and throughput. The next step is defining performance at the instance level. Each EBS Optimized instance has defined performance (both throughput and IOPS) for the set of EBS volumes attached to the instance. A customer can, therefore, size instances and volumes to get the desired level of performance. In addition, customers can use our reported metrics to observe instance level and volume level performance. They can set alarms to determine if what they are seeing does not match the expected performance – the metrics can also help determine if customers are configured at the right type of instance with the right amount of performance at the volume level or not. On the EBS end, we use the configured performance to inform how we allocate the appropriate instance and EBS infrastructure to support the volumes. By appropriately allocating infrastructure, we avoid resource contention. Additionally, we constantly monitor our infrastructure. This monitoring allows us to detect infrastructure failure (or imminent infrastructure failure) and therefore, move the volumes pro-actively to functioning hardware while the underlying infrastructure is either repaired or replaced (as appropriate).

When attached to EBS-optimized instances, General Purpose SSD (gp3 and gp2) volumes are designed to deliver within 10% of the provisioned IOPS performance 99% of the time in a given year. Your exact performance depends on your application’s I/O requirements.

When attached to EBS-optimized instances, General Purpose SSD (gp3 and gp2) volumes can achieve single digit millisecond latencies. Your exact performance depends on your application’s I/O requirements.

No. All General Purpose SSD (gp3) volumes include 3,000 IOPS and 125 MB/s of consistent performance at no additional cost. Volumes can sustain the full 3,000 IOPS and 125 MB/s indefinitely.

General Purpose SSD (gp2) volumes that are under 1,000 GB receive burst IOPS performance up to 3,000 IOPS for at least 30 min of sustained performance. Additionally, gp2 volumes deliver consistent performance of 3 IOPS per provisioned GB. For example, a 500 GB volume is capable of driving 1,500 IOPS consistently, and bursting to 3,000 IOPS for 60 minutes (3,000 IOPS * 60 seconds * 30 minutes / 1,500 IOPS / 60 seconds).

EBS Block Express is the next generation of Amazon EBS storage server architecture purpose-built to deliver the highest levels of performance with sub-millisecond latency for block storage at cloud scale. Block Express does this by using Scalable Reliable Datagrams (SRD), a high-performance lower-latency network protocol, to communicate with Nitro System-based EC2 instances. This is the same high performance and low latency network interface that is used for inter-instance communication in Elastic Fabric Adapter (EFA) for High Performance Computing (HPC) and Machine Learning (ML) workloads. Additionally, Block Express offers modular software and hardware building blocks that can be assembled in many different ways, giving us the flexibility to design and deliver improved performance and new features at a faster rate.

io2 Block Express is suited for performance and capacity intensive workloads that benefit from lower latency, higher IOPS, higher throughput, or larger capacity in a single volume. These workloads include relational and NoSQL databases such as SAP HANA, Oracle, MS SQL, PostgreSQL, MySQL, MongoDB, Cassandra, and critical business operation workloads such as SAP Business Suite, NetWeaver, Oracle eBusiness, PeopleSoft, Siebel, and ERP workloads such as Infor LN and Infor M3.

This feature can be used via the following APIs that can be called using AWS CLI or via AWS SDK.

• List Snapshot Blocks: The ListSnapshotBlocks API operation returns the block indexes and block tokens for blocks in the specified snapshot.
• List Changed Blocks: The ListChangedBlocks API operation returns the block indexes and block tokens for blocks that are different between two specified snapshots of the same volume/snapshot lineage.
• Get Snapshot Blocks: The GetSnapshotBlock API operation returns the data in a block for the specified snapshot ID, block index, and block token.
• Start Snapshot: The StartSnapshot operation starts a snapshot, either as an incremental snapshot of an existing one or as a new snapshot. The started snapshot remains in a pending state until it is completed using the CompleteSnapshot action.
• Put Snapshot Block: The PutSnapshot operation adds data in the form of individual blocks to a started snapshot that is in a pending state. You must specify a Base64-encoded SHA256 checksum for the block of data transmitted. The service validates the checksum after the transmission is completed. The request fails if the checksum computed by service doesn’t match what you specified.
• Complete Snapshot: The CompleteSnapshot operation completes a started snapshot that is in a pending state. The snapshot is then changed to a completed state.

For more information, please refer to technical documentation.

GetSnapshotBlock and PutSnapshotBlock APIs support 512KiB block size.

No, snapshots are only available through the Amazon EC2 API.

No, snapshots can be done in real time while the volume is attached and in use. However, snapshots only capture data that has been written to your Amazon EBS volume, which might exclude any data that has been locally cached by your application or OS. To ensure consistent snapshots on volumes attached to an instance, we recommend detaching the volume cleanly, issuing the snapshot command, and then reattaching the volume. For Amazon EBS volumes that serve as root devices, we recommend shutting down the machine to take a clean snapshot.

By design, an EBS Snapshot of an entire 16 TB volume should take no longer than the time it takes to snapshot an entire 1 TB volume. However, the actual time taken to create a snapshot depends on several factors including the amount of data that has changed since the last snapshot of the EBS volume.

Each snapshot is given a unique identifier, and customers can create volumes based on any of their existing snapshots.

You can find snapshots that are shared with you by selecting Private Snapshots from the list in the Snapshots section of the AWS Management Console. This section lists both snapshots that you own and snapshots that are shared with you.

You can find snapshots that are shared globally by selecting Public Snapshots from the list in the Snapshots section of the AWS Management Console. You can also restrict public access to snapshots in an account by enabling Block Public Access for EBS Snapshots.

You can use the AWS Management Console to find public datasets stored as Amazon Snapshots. Log into the console, select the Amazon EC2 Service, select Snapshots and then filter on Public Snapshots. All information on public datasets is available in our AWS Public Datasets resource center.

You should enable FSR on snapshots if you are concerned about latency of data access when you restore data from a snapshot to a volume and want to avoid the initial performance hit during initialization. FSR is intended to help with use cases such as virtual desktop infrastructure (VDI), backup & restore, test/dev volume copies, and booting from custom AMIs. By enabling FSR on your snapshot, you will see improved and predictable performance whenever you need to restore data from that snapshot.

No. FSR-enabled snapshots improve restoring backup data from your snapshot to your volumes. FSR-enabled snapshots do not speed up snapshot creation time.

To use the feature, invoke the new enable-fast-snapshot-restores API on a snapshot within the availability zone (AZ) where initialized volumes are to be restored.

The FSR-enabled snapshot may be in any one of the following states: enabling, optimizing, enabled, disabling, disabled. State transitions are published as CloudWatch events and the FSR state can be checked via the describe-fast-snapshot-restores API.

Enabling FSR on a snapshot does not change any existing snapshot API interactions, and existing workflows will not need to change. FSR can be enabled or disabled on account-owned snapshots only. FSR cannot be applied to shared snapshots. You can view the list of your FSR-enabled snapshots via API or the console.

Volumes created from an FSR-enabled snapshot are fully initialized. However, there are limits on the number of volumes that can be created with immediate full performance. These limits are expressed in the form of a credit bucket that is associated with an FSR-enabled snapshot in a given AZ. The important things to know regarding credits:

1. A single volume create operation consumes a single credit
2. The number of credits is a function of the FSR-enabled snapshot size
3. Credits refill over time
4. Maximum credit bucket size is 10

To estimate your credit bucket size and fill rate, divide 1,024 by your snapshot size. For example, a 100 GiB FSR-enabled snapshot will have the maximum balance of 10 credits with a fill rate of 10 credits every hour. A 4 TiB snapshot will have a maximum balance of 1 with a fill rate of 1 credit every 4 hours.

It's important to note that the credit bucket size is a function of the FSR-enabled snapshot size, not the size of the volumes that are created. For example, it is possible to create up to ten 1TiB volumes from a 100GiB snapshot at once.

Lastly, each AZ in which the snapshot is FSR-enabled gets its own credit bucket independent of other AZs.

The size of the create credit bucket represents the maximum number and the balance of the credit bucket represents the number of creates available. When filled, up to 10 initialized volumes can be created from an FSR-enabled snapshot at once. Both the maximum size of the credit bucket and the credit bucket balance are published as CloudWatch metrics. Volume creations beyond the limit will proceed as if FSR is not enabled on the snapshot.

When using FSR, a new EBS-specific attribute (fastRestored) is added in the DescribeVolumes API to denote the status at create time. When a volume is created from an FSR-enabled snapshot without sufficient volume-create credits, the create will succeed but the volume will not be initialized.

When you delete a snapshot, the FSR for your snapshot is automatically disabled and FSR billing for the snapshot will be terminated.

Yes, you can enable FSR for public snapshots as well as all private snapshots shared with your account. To enable FSR for shared snapshots, you can use the same set of API calls that you use for enabling FSR on snapshots you own.

When you enable FSR on your shared snapshot, you will be billed at standard FSR rates (see pricing pages). Note that only your account will be billed for the FSR of the shared snapshot. The owner of the snapshot will not get billed when you enable FSR on the shared snapshot.

When the owner of your shared snapshot deletes the snapshot, or stops sharing the snapshot with you by revoking your permissions to create volumes from this snapshot, the FSR for your shared snapshot is automatically disabled and FSR billing for the snapshot will be terminated.

You can use Amazon Data Lifecycle Manager and AWS Systems Manager (SSM) to coordinate freeze, flush I/O, and unfreeze of your application or database as well as the initialization of EBS Snapshots. You will need to provide commands to perform the actions that are specific to your application or database. You can also refer to our documentation for AWS-provided code and SSM documents for MySQL, PostgreSQL, and Windows applications.

Amazon EBS encryption offers seamless encryption of EBS data volumes, boot volumes and snapshots, eliminating the need to build and maintain a secure key management infrastructure. EBS encryption enables data at rest security by encrypting your data using Amazon-managed keys, or keys you create and manage using the AWS Key Management Service (KMS). The encryption occurs on the servers that host EC2 instances, providing encryption of data as it moves between EC2 instances and EBS storage. For more details, see Amazon EBS encryption in the Amazon EC2 User Guide.

AWS KMS is a managed service that makes it easy for you to create and control the encryption keys used to encrypt your data. AWS Key Management Service is integrated with other AWS services including Amazon EBS, Amazon S3, and Amazon Redshift, to make it simple to encrypt your data with encryption keys that you manage. AWS Key Management Service is also integrated with AWS CloudTrail to provide you with logs of all key usage to help meet your regulatory and compliance needs. To learn more about KMS, visit the AWS Key Management Service product page.

You can use Amazon EBS encryption to meet security and encryption compliance requirements for data at rest encryption in the cloud. Pairing encryption with existing IAM access control policies improves your company’s defense-in-depth strategy.

Amazon EBS encryption handles key management for you. Each newly created volume gets a unique 256-bit AES key; Volumes created from the encrypted snapshots share the key. These keys are protected by our own key management infrastructure, which implements strong logical and physical security controls to prevent unauthorized access. Your data and associated keys are encrypted using the industry-standard AES-256 algorithm.

Yes.

Yes, using customer master keys (CMKs) that are either AWS-managed or customer-managed. You can specify the volume details and encryption through a RunInstances API call with the BlockDeviceMapping parameter or through the Launch Wizard in the EC2 Console.

Yes, you can create encrypted data volume with either default or custom CMK encryption at the time of instances launch. You can specify the volume details and encryption through BlockDeviceMapping object in RunInstances API call or through Launch Wizard in EC2 Console.

Yes. See technical documentation for details.

Yes. You can share encrypted snapshots and AMIs using a customer-managed customer master key (CMK) with other AWS accounts. See technical documentation for details.

Yes, you can enable EBS encryption by default with a single setting per region. This ensures that all new volumes are always encrypted. Refer to technical documentation for more details. 

Yes, you will be billed for the IOPS provisioned when it is disconnected from an instance. When a volume is detached, we recommend you consider creating a snapshot and deleting the volume to reduce costs. For more information, see the "Underutilized Amazon EBS Volumes" cost optimization check in Trusted Advisor. This item checks your Amazon Elastic Block Store (Amazon EBS) volume configurations and warns when volumes appear to be underused.

Except as otherwise noted, our prices are exclusive of applicable taxes and duties, including VAT and applicable sales tax. For customers with a Japanese billing address, use of AWS services is subject to Japanese Consumption Tax. Learn more.

No. Multi-Attach can be enabled on an EBS Provisioned IOPS volume and there will be charges for the storage (GB-Mo) and IOPS (IOPS-Mo) provisioned.

No.

The volume's deleteOnTermination behavior is determined by the configuration of the last attached instance that is terminated. To ensure predictable delete on termination behavior, enable or disable 'deleteOnTermination' for all of the instances to which the volume is attached.

If you want the volume to be deleted when the attached instances are terminated, enable ‘deleteOnTermination’ for all of instances to which the volume is attached. If you want to retain the volume after the attached instances have been terminated, disable ‘deleteOnTermination’ for all attached instances. For more information, see Multi-Attach technical documentation.

Your application can use Multi-Attach if the application is built on Windows Server Failover Cluster, coordinates safe access to shared storage using NVMe reservations, or coordinates safe access at the application-level.