With most instance types can use EBS as the volume storage mechanisms as the instance store volumes are not supported on all instance types. We can also use EBS when the instance store volume assigned to the instance type does not satisfy our requirements because of the following:

• Persistent storage that is highly available 
• Storage that will survive reboots and instance failure
• Block volume sizes up to 16 TB

EBS volumes are delivered from a durable, block-level storage device and can be attached to one EC2 instance at a time. The EC2 instance that has the volume attached is able to store data persistently. Once stored, the data will survive any EC2 life cycles or EC2 instance failures. We also have the ability to detach the volume from the existing EC2 instance and attach it to another one. This makes it very simple to perform updates of the operating system with the ability to roll back in case of issues.

For example, a database server running on an EC2 instance stores the data to an EBS volume. When the need arises to update the operating system or the database platform, we can simply deploy a new instance with an updated operating system and an updated database platform. We transfer the configuration from the existing EC2 instance to the new one, power the existing instance down, and detach the volume. Next, we attach the volume to the newly created, updated EC2 instance and power it on. The instance is able to access the data on the attached volume and spin up the database since we have already transferred the configuration from the old one. We can now perform some tests on the newer instance and put it in production or revert to the old instance by powering the new one down, attaching the volume to the old one, and powering the old instance back on.

There are several different types of EBS volumes that we can use with our EC2 instances:

• General Purpose SSD (gp2): General purpose SSD-backed disks with up to 10,000 IOPS, 160 MBps throughput, and a size of between 1 GB and 16 TB per volume
• Provisioned IOPS SSD (io1): Performance optimized SSD-backed disks with up to 32,000 IOPS, 500 MBps throughput, and a size of between 4 GB and 16 TB per volume
• Throughput Optimized HDD (st1): Throughput optimized magnetic disks with up to 500 IOPS, 500 MBps throughput, and a size of between 500 GB and 16 TB per volume
• Cold HDD (sc1): Lowest cost magnetic disks with up to 250 IOPS, 250 MBps throughput, and a size of between 500 GB and 16 TB per volume

If the limitations of EBS volumes do not suffice, you have the ability to attach multiple volumes and create a Redundant Array of Independent Disks (RAIDs) across multiple volumes, although certain limitations do apply, for instance the maximum IOPS per instance using EBS volumes is limited to 80,000 and the maximum throughput per instance is 1,750 MBps.

EBS volumes are also designed to be highly available within an availability zone and adhere to the EC2 99.99% SLA. Each volume is replicated to two different physical devices within the same availability zone, so any disk failure or failure of the backend disk subsystem will not affect the EBS volume. Even though the EBS volumes are highly available within an availability zone, failures can occur that will bring the availability zone down and the EBS volumes with it. Also, there is no guarantee for data durability on an EBS volume, so make sure you protect the volumes with important data by either replicating the data to another instance in another availability zone with another EBS volume attached or snapshot the EBS volume.