AWS Elastic Compute Cloud (EC2) - Quick Notes

AWS Elastic Compute Cloud

AWS Elastic Compute Cloud, also referred to as EC2, is a cloud service that provides a secure and auto scalable compute capacity in the cloud.

The following is the summary of the various features/capabilities of EC2:

• Provides virtual computing environment known as Instances

• Reduces time to provision and launch new instances

• Allows one to quickly scale capacity up or down

• Pay only for the capacity used

• Allows one to build failure resilient applications

• Instance(s) are provided as preconfigured templates and are called Amazon Machine Images (or AMIs for short)

• AMI is a virtual image that packages the desired operating system and necessary software package(s) that is Region specific

• When an AMI is copied from one Region to a second Region, it automatically creates a Snapshot in the second Region

• Provides different configuration of instances that are optimized for CPU, memory, storage, and networking and are known as Instance Types

• Allows one to securely login to instances using asymmetric keys known as Key Pairs

• Storage volumes called Elastic Block Store (or EBS for short) can be attached to instances for storing data

• Provides a virtual firewall called Security Group that enables one to specify the protocols, ports, and source IP ranges that can reach an instance

EC2 Instance Types

AWS provides a large variety of EC2 Instance Types to choose from based on one's workload needs.

EC2 provides the following types of instances:

Type	Description
General Purpose	Provides a balance of compute, memory and networking resources, that can be used for a variety of general workloads
Compute Optimized	Suitable for compute bound applications that benefit from high performance processors, that can be used for batch processing, media transcoding, high performance computing (HPC), machine learning, etc
Memory Optimized	Delivers fast performance for workloads that process large data sets in memory
Storage Optimized	Designed for workloads that demand fast access, low latency I/O operations for processing large data sets
Accelerated Computing	Make use hardware accelerators or graphical processing units (GPUs), that can be used for demanding workloads such as high-performance computing (HPC), artificial intelligence, etc

EC2 Purchasing Options

EC2 provides the following Purchasing Options for Instances:

• Pay a fixed price by the second

• No long-term commitments or up-front costs

• Better for short-term, irregular, uninterrupted workloads

• Upto 72% savings on the On-Demand price

• One or three year commitment

• Commitment to a consistent amount of usage (in USD per hour)

• Commitment to a consistent instance configuration, such as, instance type, Region, Tenancy, Platform

• One or three year commitment

• Significantly lower hourly rate

• Better for applications with steady predictable workloads

• Upto 90 percent savings on the On-Demand price

• Allows one to bid on unused EC2 computing capacity at steep discounts

• Bid by specifying a maximum price willing to pay. Instance is launched when the spot price is lower than the specified maximum price

• Will continue to run until terminated or the spot price exceeds the specified maximum price

• If the spot price exceeds the specified maximum price, the instance will receive a two-minute notification before it is terminated

• Better for applications that have flexible start and end times

• Physical EC2 server for dedicated use

• Allows one to use existing per-socket, per-core, or per-VM software licenses

• Helps in cases where there is a need to meet compliance and regulatory requirements

• Run on hardware that is dedicated to a single customer

• May share hardware with other instances from the same AWS account

EC2 Fleet and Spot Fleet

EC2 Fleet and Spot Fleet allow a user to launch a group of compute Instances on AWS.

The following are some of the features:

• Launch multiple Instance Types

• Take advantage of a mix of On-Demand, Reserved Instances, Savings Plan, and Spot Instances

• Allows one to optimize costs through the use of Spot Instances

• Automatically distributes instances evenly across multiple Availability Zones within a Region for resiliency

• Automatically request replacement Spot Instances if they are interrupted

Elastic IP Address

Elastic IP is a public static IPv4 address that can be assigned to a single Instance.

The following are some of the features:

• Remains static and does not change over time

• Incurs an hourly charge whether it is used or not

• One can mask the failure of an Instance by rapidly remapping it to another Instance

• For use in a specific Region only

• All AWS accounts have a quota of FIVE (5) Elastic IP addresses per Region

Placement Groups

For certain workloads that have a need for a group of interdependent Instances, one can create a Placement Group and attach it to Instances to influence their placement.

The following are the three types of Placement Groups:

• Packs all the Instances in a single Availability Zone

• For workloads that have a low-latency network performance requirement

• Distribute a group of Instances that are placed on different racks in an Availability Zone

• Racks can also be on different Availability Zones in the same Region

• Spreads Instances across different logical segments called Paritions (in different racks) in multiple Availability Zones in the same Region

• There can be a maximum of SEVEN (7) Partitions per Availability Zone

• Number of Instances that can be launched in a Partition is limited to the limits of the AWS account

Elastic Network Interfaces

Elastic Network Interface is a logical networking component and represents a virtual network interface card.

The following are some of the features:

• Each Instance is assigned a default Elastic Network Interface, referred to as the Primary Network Interface and cannot be detached from an Instance

• One can create and attach additional secondary Elastic Network Interfaces to an Instance

• Can have a primary private IPv4 and one or more secondary private IPv4

• Can associate one Elastic IP address per private IPv4 address

• Can associate one public IPv4 address

• All IPv4 addresses must come from the Subnet it resides in

• Bound to a Subnet in an Availability Zone in a Region

Security Groups

A Security Group behaves like a virtual firewall in EC2 Instances and controls incoming and outgoing network traffic.

The following are some of the features:

• Are Region specific

• Is bound to the ENI of the Instance

• An Instance can have one or more Security Groups associated with it

• By default all inbound traffic is blocked while all outbound traffic is allowed

• Rules are only of type ALLOW

• Rules enable one to filter traffic based on protocols and port numbers

• Rules can be all added or removed at any time and they take effect immediately

• For Instances with multiple Security Groups, all the rules are aggregated before being applied

EC2 Instance Store

EC2 Instance Store is an ephemeral, high-performance physical hardware storage disk which can be attached to an Instance.

The following are some of the features:

• Ideal for temporary storage of information that changes frequently, such as caches, scratch data, and other temporary content

• Size as well as the number of devices available varies by Instance Type

• Can be attached only at Instance launch and NOT after launch

• Exists only during the lifetime of the Instance

• Data does not persist if the Instance is stopped, hibernated, or terminated

References

Official AWS EC2 Documentation

Official AWS EC2 Instance Types

Official AWS EC2 FAQs

© PolarSPARC