Table of Contents

• Reference
• How-to
• Topics
  • EBS-Optimized Instances
  • Monitor I/O Characteristics with CloudWatch
  • gp2: I/O Credits and Burst Performance
  • io1: For the best per-I/O latency experience
  • io1: Within 10 percent of the provisioned IOPS performance?
• Links

Reference

How-to

Topics

EBS-Optimized Instances

• To ensure sustained reliable network bandwidth between your EC2 instance and your EBS volumes:
  • To launch the EC2 instance as EBS-optimized
  • To choose an instance type with *10 Gigabit network connectivity*(like r4.8xtype).

Be sure to choose an EBS-optimized instance that provides more dedicated EBS throughput than your application needs; otherwise, the Amazon EBS to Amazon EC2 connection will become a performance bottleneck.

When EBS-Optimized, Each instance can acquire IO Performance following:

Monitor I/O Characteristics with CloudWatch

• BurstBalance: burst bucket balance for gp2, st1, and sc1
• VolumeQueueLength: To make sure your application is not trying to drive more IOPS than you have provisioned
• VolumeReadBytes
• VolumeWriteBytes
• VolumeReadOps
• VolumeWriteOps

gp2: I/O Credits and Burst Performance

io1: For the best per-I/O latency experience

We recommend that you provision an IOPS-to-GiB ratio greater than 2:1. For example, a 2,000 IOPS volume should be smaller than 1,000 GiB.

io1: Within 10 percent of the provisioned IOPS performance?

• Ambiguous wording

if you provision a volume with 5,000 IOPS, Amazon EBS delivers an average of 4,500 IOPS or higher for 99.9 percent of the time over a year. I think the document's wrong. They should edit it like "within 90%".

It's more like this: 99.9% of the time over the year, AWS will deliver within 4000 IOPS +(?)/- 10%.

Links