Skip to content

Latest commit

 

History

History
41 lines (29 loc) · 1.53 KB

ec2-placement-groups.md

File metadata and controls

41 lines (29 loc) · 1.53 KB

EC2 Placement Groups

Three types of placement groups

Logically group EC2 instances

Cluster Placement Group

  • Grouping of instances within a single availablity zone
  • Recommended for applications that need low network latency, high network throughput, or both (HPC)
  • Only certain instance types can be launched into a cluster group
  • Cluster placement groups can't span mulitple availablity zones

Spread Placement Group

  • Group of instances placed on distinct underlying hardware
  • Recommended for applications that have a small number of critical instances that should be kept seperate from each other
  • Use for individual instances
  • Spread placement groups can span multiple availablity zones

Partition Placement Group

  • Each partion placement group has it's own set of server rack
  • Each rack has it's own network and power source
  • No two partions share the same rack, allow you to isolate the impact of a hardware failure within a application
  • EC2 logically divides each group into logical segments called partitions
  • Used for multiple instance with dedicated network and power sources

Other notes

  • Only certain instance types can be launched into a placement group

    • Compute optimized
    • GPU
    • Memory optimized
    • Storage optimized
  • Placement groups can't be merged

  • AWS recommneds homogenous instances within a cluster placement group

  • Existing instances can be moved into a placement group if they are in a stopped state

  • Instance in placement groups can be removed via the AWS CLI or SDK (not console)