2. Amazon Elastic Compute Clouds (EC2)

IP Addresses

Placement Groups

Elastic Network Interfaces

NAT Gateway vs NAT Instances

Instance Lifecycle

Billing and Provisioning Options

AWS Nitro System and Enclaves

Misc

Explore

Placement Groups

Placement groups in AWS provide a way to logically group instances to meet specific application requirements, such as low latency, high throughput, or isolation. Here's an overview of the three types of placement groups available:

Cluster Placement Group:

Description: Groups instances into a single Availability Zone (AZ) to achieve low-latency communication and high network throughput.

Use Cases: Ideal for applications with high inter-instance communication requirements, such as HPC (High-Performance Computing) and distributed data analysis.

Recommendation: Suitable when most network traffic is between instances within the group.

Advantages: Low latency, high throughput within the placement group.

⁠

Partition Placement Group:

Description: Divides instances into logical segments called partitions, with each partition having its own set of racks, network, and power sources.

Use Cases: Designed for large distributed and replicated workloads, such as HDFS (Hadoop Distributed File System), HBase, and Cassandra.

Recommendation: Deploying distributed and replicated workloads across distinct racks.

Advantages: Isolation of hardware failures within the application, enhanced fault tolerance for distributed workloads.

⁠

Spread Placement Group:

Description: Spreads instances across distinct underlying hardware to minimize the impact of hardware failures.

Use Cases: Suitable for critical applications where instances need to be kept separate to enhance fault tolerance.

Recommendation: Used when a small number of critical instances should be isolated from each other.

Advantages: Enhanced fault tolerance, isolation of critical instances.

⁠

Differences between clustered and spread placement groups

Differences between clustered and spread placement groups

Column 1

Clustered

Partition

Spread

What

Instances are placed into a low-latency group within a single AZ

Instances are grouped into logical segments called partitions which use distinct hardware

Instances are spread across underlying hardware

When

Need low network latency and/or high network throughput

Need control and visibility into instance placement

Reduce the risk of simultaneous instance failure if underlying hardware fails

Pros

Get the most out of enhanced networking Instances

Reduces likelihood of correlated failures for large workloads.

Can span multiple AZs

Cons

Finite capacity: recommend launching all you might need up front

Partition placement groups are not supported for Dedicated Hosts

Maximum of 7 instances running per group, per AZ

There are no rows in this table

⁠

Spread Placement Groups: Enhancing Fault Tolerance and Performance

Spread placement groups in AWS offer a way to distribute instances across separate underlying hardware, minimizing the risk of simultaneous failures and enhancing performance for latency-sensitive applications. Here's a closer look at the key aspects and recommendations for using spread placement groups effectively:

Risk Reduction: By spreading instances across distinct hardware, spread placement groups mitigate the risk of simultaneous failures that might occur when instances share the same underlying hardware.

Low Latency, High Bandwidth: Spread placement groups are recommended for applications that benefit from low latency and high bandwidth, such as real-time analytics and sensitive transactional systems.

Enhanced Networking: It's recommended to use instance types that support enhanced networking capabilities for optimal performance within spread placement groups.

Communication Modes: Instances within a spread placement group can communicate with each other using either private or public IP addresses.

Performance Considerations:

Private IP Addresses: Best performance is achieved when instances communicate using private IP addresses within the VPC.

Public IP Addresses: Using public IP addresses may limit performance to 5 Gbps or less due to potential network constraints.

Network Speed: Spread placement groups offer low-latency network connectivity with speeds of 10 Gbps or 25 Gbps, enabling high-speed data transfer between instances.

Homogeneous Instance Types: It's recommended to keep instance types homogeneous within a spread placement group to ensure consistent performance characteristics and simplify management.

Reserved Instances: While you can use reserved instances at the instance level, it's important to note that you cannot reserve capacity explicitly for a placement group. Capacity reservations are typically at the Availability Zone level.

Naming and Unique Identifier: The name specified for a placement group must be unique within your AWS account for the region where it's created.

Limitations:

Merging: Placement groups cannot be merged, so plan your placement group strategy carefully.

Single Group: An instance can be launched in only one placement group at a time and cannot span multiple placement groups.

Tenancy: Instances with a tenancy of "host" cannot be launched in placement groups.

Want to print your doc?
This is not the way.

Try clicking the ⋯ next to your doc name or using a keyboard shortcut (

CtrlP

) instead.