3.29 Instances lab AWS

This lab walks through the EC2 launch types we covered theoretically in the previous lesson. We open the AWS console and explore the reserved instances catalog. Choosing for example an m4.large with a 36-month term and zero upfront payment, we can simulate the price for four machines: about $2,442 over 36 months plus a $0.023/hour usage rate. We don't actually confirm — the goal is just to walk the workflow.

What we explore in the console

  • Scheduled instances — only available in some regions (e.g. US East N. Virginia). We pick a start date, a 24-hour daily window, a weekly recurrence on Sunday, and an instance type.
  • Dedicated hosts — bring-your-own-license model. We pick the instance family m5.4xlarge in us-east-3a then Allocate host.
  • Spot requests — we view the historical pricing chart, define a workflow (load balancing, big data) and a maximum bid.
  • Spot blocks — guarantee 1h to 6h of uninterrupted runtime, useful when partial reclaim would be unacceptable.

For the spot request creation form, we pick a balanced workflow, select an instance type, the VPC, availability zones, the key pair, the requested capacity (vCPUs or number of instances). We also choose whether to maintain target capacity — the spot fleet then tries to keep the requested number running at all times. The summary shows a recap of the fleet composition, the allocation strategy (lowest price, diversified, or capacity-optimized) and the hourly cost compared to on-demand.

Finally we launch a standalone spot instance from the regular Launch instance wizard: pick the Amazon Linux 2 AMI, a t2.micro, then in the network step toggle Request Spot instances. Set the maximum hourly price, the interruption behavior (terminate / stop / hibernate), the request validity, the number of spot instances. The rest of the procedure is identical to launching an on-demand instance. You now know how to launch every EC2 type: on-demand, reserved, scheduled, spot, spot fleet, and dedicated host.

Summary

This AWS lab lesson explores EC2 instance purchasing models beyond on-demand pricing. It covers reserved instances for long-term commitments (12-36 months) with lower hourly costs, scheduled instances for time-specific deployments with configurable recurrence, and BYOL (Bring Your Own License) instances for regulatory compliance and cost optimization. The lesson concludes with spot instances and spot requests, demonstrating how allocation strategies (capacity-optimized, diversified, lowest-price) enable significant cost savings while maintaining capacity requirements.

Key points

  • Reserved instances require 12-36 month commitments and offer lower hourly rates; you can choose upfront payment levels (zero, partial, or full) to balance cost and cash flow
  • Scheduled instances allow you to plan instance creation for specific dates and times with recurring schedules (daily, weekly, monthly); availability is regional and not all regions support them (e.g., unavailable in Paris, available in North Virginia and Ireland)
  • BYOL (Bring Your Own License) instances enable cost reduction by using existing licenses and support regulatory compliance requirements for enterprises
  • Spot instance requests allow automated deployment with dynamic pricing; you can view pricing history and set maximum prices to control costs
  • Spot allocation strategies include capacity-optimized (prioritizes availability), diversified (spreads across multiple instance types), and lowest-price (selects cheapest options across zones)
  • Spot instances typically cost significantly less than on-demand (example shows under $0.10/hour for micro instance) but can be interrupted if your maximum price threshold is exceeded

FAQ

What are the key differences between reserved instances and scheduled instances?

Reserved instances require a 12-36 month commitment with a fixed hourly rate and optional upfront payment, providing guaranteed capacity. Scheduled instances are planned for specific dates/times with configurable recurrence but are only available in certain regions and offer different pricing flexibility.

How does BYOL (Bring Your Own License) help reduce AWS costs?

BYOL allows you to use software licenses you already own rather than paying AWS licensing costs. This reduces overall expenses and helps organizations meet regulatory and compliance requirements by using licenses they have already purchased.

What allocation strategies are available for spot instances, and how do they differ?

Spot requests support three allocation strategies: capacity-optimized (maintains required capacity first), diversified (spreads instances across multiple instance types and zones for reliability), and lowest-price (selects the cheapest instances available). The choice depends on whether you prioritize cost savings or capacity stability.