Skip to content

Latest commit

 

History

History
41 lines (37 loc) · 1.7 KB

HPC_Placement_Groups.md

File metadata and controls

41 lines (37 loc) · 1.7 KB

HPC and Placement Groups

  • Return to table of contents

  • Useful Links:

  • Notes:

    • Cluster placement group:
      • Single AZ:
        • Cannot not span AZs.
        • Can span peered VPC, but impacts performance.
        • Maximum possible performance between EC2 instances.
        • Benefits:
          • Highest levels of throughput.
          • Lowest levels of latency.
        • Capacity gets allocated in advanced.
        • Use same type and launch time of EC2 instances.
      • Partition placement group:
        • EC2 attempts to distribute instances equally across partitions.
        • Large scale, replicated workloads where resilience is need.
        • 7 instances per AZ (HARD LIMIT).
        • Great for topology aware applications:
          • HDFS, HBase, Cassandra.
      • Spread placement group:
        • Small set of infrastructure where you need highest level of availability.
          • Each instances rune from a different rack.
          • Each rack has its own network and power source.
        • Single AZ or multi AZ.
        • 7 instances per AZ (HARD LIMIT)

  • Exam Tips:

    • Cluster placement group:
      • Cannot really modify a placement group, have difficulty adding instances.
      • Better outcome when you provision the same type of instance.
      • Performance vs resilience? Cluster for performance.
      • Can span VPC peers, but do lose some benefits of extra bandwidth.
    • Partition placement group:
      • Large scale, replicated workloads where resilience is need.
    • Spread placement group:
      • Small set of infrastructure where you need highest level of availability.