AWS recently introduced a new provisioned IOPS volume type (io2) for high-performance databases and workloads that offers a durability of 99.999% and the ability to provision up to 500 IOPS for every GiB of storage.
Jeff Barr, chief evangelist at AWS, explains the main benefits of the new EBS volume type:
The io2 volumes are designed to deliver 99.999% durability, making them 2000x more reliable than a commodity disk drive, further reducing the possibility of a storage volume failure and helping to improve the availability of your application. (...) You can get higher performance from your EBS volumes, and you can reduce or outright eliminate any over-provisioning that you might have done in the past to achieve the desired level of performance.
There are currently six types of EBS storage available on AWS that differ in performance characteristics and price, and io2 is the first increase in IOPS to storage in over four years. The new EBS volumes are available in 16 regions and provide a 100x higher durability and 10x more IOPS/GiB compared to the previous generation io1 volumes.
There is one significant limitation, compared to the existing provisioned IOPS volumes: the new io2 volumes do not support multi-attach, the ability to attach a volume to multiple EC2 instances to achieve higher availability.
Luc van Donkersgoed, head of AWS technology at Sentia Group, performed a detailed comparison of the io1 and io2 volumes for different workloads and suggests to take advantage of the new ones for any new deployment:
Smaller volumes are up to 3.33% cheaper on io2 than on io1. All EBS volumes before the release of the io2 type have a durability of 99.8 - 99.9% on a yearly basis. That translates to 1 or 2 failing drives per 1.000 volumes, every year. The new io2 disk type features a 99.999% reliability - which means one volume in every 100,000 volumes will fail on a yearly basis, bringing io2 disks to enterprise levels of reliability. (...) With the cost savings and durability improvements in io2, there is no reason not to migrate as many io1 disks to io2 as soon as you can.
Nafea Bshara, VP and distinguished engineer at AWS, commented:
Very rare to see this level of technology leapfrog, especially in enterprise-grade storage (...) helping AWS customers migrate their on-prem SAN workloads, gain in durability, performance and reduce cost.
Amazon is not the first cloud provider offering over 100 IOPS/GiB and 64K IOPS per disk. Azure ultra disks support IOPS limits of 300 IOPS/GiB, with up to a maximum of 160K IOPS per disk. Google Cloud SSD persistent disks support up to 30 write IOPS per GiB, with read IOPS up to 30K and write IOPS up to 100K per instance.
