AWS Module 6 -- Section 3

Amazon EC2 Pricing Models

On-Demand Instances

Spot Instances

Reserved Instances

Dedicated Hosts

Dedicated Instances

Scheduled Reserved Instances
Per Second Billing

On-Demand Instances

• Pay by the hour

• No long term commitments 

  • Lowest upfront cost

  • Most flexibility

• Eligible for the AWS Free Tier

  • Good choice for applications with short-term, spiky, or unpredictable workloads

Spot Instances

• Instances run as long as they are available and your bid is above the Spot Instance price

• They can be interrupted by AWS with a 2-minute notification

• Interruption options include terminated, stopped, or hibernated

• Prices can be significantly less expensive compared to On-Demand instances

  • Good choice when you have flexibility in when your applications can run (not a specific time)

Reserved Instances

• Full, partial, or no upfront payment for Instance you reserve

• Discount on fixed hourly charge for that instance

  • 1 year or 3 year term

• Should be used if you expect consistent, heavy use, in order to get substantial savings compared to On-Demand Instances (predictable)

Dedicated Hosts

• A physical server with EC2 instances capacity fully dedicated to your use

  • Enable you to use you existing per-socket, per-core, or per-VM software licenses

  • Enable you to use licenses for Microsoft Windows or Microsoft SQL servers

Dedicated Instances

• Instances that run in a VPC on hardware that is dedicated to a single customer

• Physically isolated at the host hardware level

Scheduled Reserved Instances

• Purchase a capacity reservation that is always available on a recurring schedule you specify (daily, weekly or monthly)

• 1 year term

• You pay for the time that the instances are scheduled (even if you are not actively using it)

Per Second Billing

• Available for On-Demand instances, Reserved Instances, and Spot instances that run Amazon Linux or Ubuntu

The Benfits of EC2 Pricing Models - On-Demand Instances

On-Demand Instances - 

Offers the most flexibility, with no long-term contracts and low costs

The Benfits of EC2 Pricing Models - Spot Instances

Spot Instance - 

Provide large scale, dynamic workloads at a significantly discounted price

The Benfits of EC2 Pricing Models - Reserved Instances

Reserved Instances - 

Good choice if there is a predictable or steady-state compute need

• Predictability ensures compute capacity is available when needed

(Example: Instance that you know you want to keep running most or all of the time for months or years)

The Benfits of EC2 Pricing Models - Dedicated Hosts

Dedicated Hosts - 

Good choice when you have licensing restrictions for the software you want to run on Amazon EC2, or when you have specific compliance or regulatory requirements that preclude you from using the other deployment options

Amazon EC2 Pricing Models: Use cases - On-Demand Instances

On-Demand Instances - 

• Short-term, spiky, or unpredictable workloads

• Application development or testing

Amazon EC2 Pricing Models: Use cases - Spot Instances

Spot Instance - 

• Applications with flexible start and end times

• Applications only feasible at very low compute prices

• Users with urgent computing needs for large amounts of additional capacity

Are instances terminated by default?

What common use cases can be used to combat this?

Yes

Web servers, API backends, and big data processing

Amazon S3 —> Constantly saves data for persistent storage

Amazon EC2 Pricing Models: Use cases - Reserved Instances

Reserved Instances - 

• Steady state or predictable usage workloads

  • Long term workloads that will run in a consistent way over many months

• Applications that require reserved capacity, including disaster recovery

• Users able to make upfront payments to reduce total computing costs even further

Amazon EC2 Pricing Models: Use cases - Dedicated Hosts

Dedicated Hosts - 

• Bring your own license (BYOL)

• Compliance and regulatory restrictions

• Usage and licensing tracking

• Control Instance placement

What are the 4 pillars of Cost Optimization?

1. Right-size – Choose the right balance of instance types 

• Servers can be sized down or turned off and still provide performance
  

2. Increase elasticity – Design your deployments to reduce the amount of server capacity that is idle by implementing deployments that are elastic

• Deployments that use automatic scaling to handle peak loads
  

3. Optimal pricing model – 

• Analyze your usage patterns so that you can run EC2 instances
  

4. Optimize storage choices – Analyze the storage requirements of your deployments

• Reduce unused storage overhead when possible, and choose less expensive storage options if they can 

Pillar 1: Right Size

• Provision Instances to match the need

  • CPU, memory, storage, and network throughput

  • Select appropriate instance types for your use

• Use Amazon CloudWatch metrics

  • How idle are instances? When?

  • Downsize instances

• Best practice: Right Size, then Reserve

How many instances types and size are there?

60

In order to Right Size, what must one do?

1) Select the cheapest instance available (still needs to meet performance requirements)

2) Review CPU, RAM, storage, and network utilization to identify instances that COULD be downsized

3) Use Amazon CloudWatch metrics and set up custom metrics

Pillar 2: Increase elasticity

• Stop or hibernate Amazon EBS-backed instances that are not actively in use

  • Example: non-production development or test instances

• Use automatic scaling to match needs based on usage

  • Automated and time-based elasticity

Pillar 3: Optimal Pricing Model

• Leverage the right pricing model for your use case

  • Consider your usage patterns

• Optimize and combine purchase types

  • Examples:

    • Use On-Demand Instance and Spot Instance for variable workloads

    • Use Reserved Instances for predictable workloads

• Consider serverless solutions (AWS Lambda)

Pillar 4: Optimize Storage Choice

• Best Practice: Reduce costs while maintaining storage performance and availability

• Resize EBS volumes

• Change EBS volume types

  • Can you meet performance requirements with less expensive storage?

  • Example: Amazon EBS Throughput Optimized HDD storage typically costs half as much as the default General Purpose SSD (gp2) storage option

• Delete EBS snapshots (backups) that are no longer needed

• Identify the most appropriate destination for specific types of data

  • Does the application need the instance to reside on Amazon EBS?

  • Amazon S3 storage options with lifecycle policies can reduce costs

What is the motto for the recommendations for Cost Optimization?

What type of Process is Cost Optimization?

Measure, monitor and improve

Type —> Ongoing

What are the recommendations for Cost Optimization?

• Recommendations –

  • Define and enforce cost allocation tagging

    • Utilize Billing and Cost management console

  • Define metrics, set targets, and review regularly

  • Encourage teams to architect for cost

    • Utilize AWS Cost Explorer

  • Assign the responsibility of optimization to an individual or to a team