I’m taking AWS training from Linux Academy along with some peers. These are continuations of my notes.

Wow it’s been almost a month since my last training notes? Time flies!

I have fired up a few EC2 spot instances as temporary labs for a few hours each, so I’m feeling pretty good about this topic. But it sure is long, so maybe I’m missing a few things.

AWS Certified Solutions Architect - EC2

Linux Academy’s EC2 section notes.

EC2 Essentials

  • EC2 Classic - I won’t have an opportunity to use it in my current role or home lab, but I’ll need to pay attention to the training as it may be on the test, and I may run into it if I go to other companies
  • Ok, I do need to learn better the different tiers as “cheapest spot instance” won’t always be my selection criteria

    • T2 - Burstable performance
    • M3 - Balanced (fair?)
    • C4 - Compute optimized
    • R2 - Memory optimized
    • G2 - GPU optimized
    • I2 - Storage optimized
    • EBS Optimized (wut? This is its own bullet point in the training verbatim; not sure where the label ends and the description begins)

      • Instructure says it has higher network priority with the load balancing
    • small/medium/large have different network throughput and network interface capabilities
  • Storage

    • Instance storage - ephemeral

      • Erased if “shut down” or “stopped” (knew this)
      • But can survive reboot (didn’t know this)
      • Underlying storage is physically attached to instance hardware (didn’t know that)
    • EBS volumes - network-attached storage; durable

      • IOPS and how data is chunked is discussed; want to study these as they sound like question fodder (256kb chunks)

        • Can provision up to 20k IOPS
        • But throughput also affected by EC2 instance since/optimization
      • Volume types

        • General Purpose SSD

          • 3 IOPS/GiB, but burstable/accruable (e.g. 1 GiB of drive space gets 3 IOPS / 256k-chunks per second)
          • Good for burstable throughput
          • Volume sizes 1Gib - 16TiB per slide, but thought instructure said “petabytes” in audio
        • Provisioned IOPS

          • IOPS as provisioned size
          • Sustained heavy use
          • 4GiB to 16TiB per visual but instructor sounded like he said “16 gigabytes”
          • Can provision up to 20k IOPS
        • Magnetic - cheap, lower work capacity, 1GiB - 1024GiB
        • Pre-warming volumes: Can increase performance by touching all blocks after provisioning
        • Snapshots - he gave a confusing explanation of snapshots, deleting old snapshots, and “data still there” even though snapshots are incremental. But I think I know how snapshots work.

          • Emphasized heavily: snapshots are stored on “S3 storage volumes”, but don’t have the ability to go to that bucket and list them
  • EC2 - start/stop/terminate, IAM, not paying for compute time when instance down but paying for storage, termination protection, cli/sdk, user-data/cloud-init

    • From within instance, http://169.254.169.254/latest/meta-data
    • http://169.254.169.254/latest/user-data
    • Can use the above for automation / discovery / registration
  • ELB

    • Can act as private lb, too, within VPC
  • EC2-Classic

    • Still important
    • Instances are assigned public IP and a cname
    • Each instance gets private IP but NOT part of a VPC
    • Private IP will change after instance shut down and brought back up
  • Security Groups

    • Say that best practice is to also have software firewall (um, ok)
    • Instance can belong to multiple security groups (? I didn’t know this)
    • Security groups can reference self as “source” in rules

Understanding Reserved Instances

(to be continued…)