Predictive scaling overview in AWS - Commands & Configuration

Introduction

Predictive scaling helps your cloud servers prepare for future traffic by learning from past patterns. It automatically adjusts the number of servers before demand changes, so your app stays fast and reliable without wasting resources.

When your website traffic grows and shrinks in a regular daily or weekly pattern.

When you want to avoid slowdowns during expected busy times like sales or events.

When you want to save money by not running too many servers during quiet periods.

When manual scaling is too slow or error-prone for your app's needs.

When you want your cloud to adjust automatically based on learned trends.

Commands

This command schedules a predictive scaling action for the Auto Scaling group named 'my-app-asg'. It sets the minimum, maximum, and desired number of servers based on expected traffic patterns for the first week of June, recurring daily at midnight.

Terminal

aws autoscaling put-scheduled-update-group-action --auto-scaling-group-name my-app-asg --scheduled-action-name predictive-scaling --start-time 2024-06-01T00:00:00Z --end-time 2024-06-07T23:59:59Z --recurrence "0 0 * * *" --min-size 2 --max-size 10 --desired-capacity 5

Expected Output

No output (command runs silently)

→

--auto-scaling-group-name - Specifies the name of the Auto Scaling group to apply predictive scaling.

→

--scheduled-action-name - Names the scheduled scaling action for easy identification.

→

--recurrence - Defines how often the scaling action repeats using cron syntax.

This command checks the scheduled scaling actions for the Auto Scaling group to confirm that the predictive scaling schedule is set correctly.

Terminal

aws autoscaling describe-scheduled-actions --auto-scaling-group-name my-app-asg

Expected Output

ScheduledUpdateGroupActions:

- AutoScalingGroupName: my-app-asg

  ScheduledActionName: predictive-scaling

  StartTime: 2024-06-01T00:00:00Z

  EndTime: 2024-06-07T23:59:59Z

  Recurrence: 0 0 * * *

  MinSize: 2

  MaxSize: 10

  DesiredCapacity: 5

→

--auto-scaling-group-name - Filters the scheduled actions for the specified Auto Scaling group.

Key Concept

If you remember nothing else from this pattern, remember: predictive scaling uses past traffic patterns to automatically adjust your server count ahead of time, keeping your app ready and cost-efficient.

Common Mistakes

Setting the min-size and max-size too close together.

This limits the flexibility of scaling and can cause performance issues during unexpected traffic spikes.

Set a reasonable range between min-size and max-size to allow scaling to adjust smoothly.

Not verifying the scheduled actions after creation.

You might think predictive scaling is active when it is not, leading to no automatic scaling.

Always run describe-scheduled-actions to confirm your scaling schedule is correctly applied.

Summary

Use 'aws autoscaling put-scheduled-update-group-action' to create predictive scaling schedules.

Check your schedules with 'aws autoscaling describe-scheduled-actions' to ensure they are active.

Set appropriate min, max, and desired server counts to allow smooth scaling based on predicted traffic.

Practice

1. What is the main purpose of AWS Predictive Scaling?

easy

A. To prepare servers ahead of traffic changes using machine learning

B. To manually add servers when traffic increases

C. To reduce server count only after traffic drops

D. To monitor server health without changing capacity

Solution

1. Step 1: Understand Predictive Scaling's goal

   Predictive Scaling uses machine learning to forecast traffic and adjust server capacity before changes happen.

2. Step 2: Compare options to this goal

   Only To prepare servers ahead of traffic changes using machine learning describes preparing servers ahead of traffic changes using machine learning.

3. Final Answer:

   To prepare servers ahead of traffic changes using machine learning -> Option A

4. Quick Check:

   Predictive Scaling = Prepare servers ahead [OK]

Hint: Predictive means it acts before traffic changes [OK]

Common Mistakes:

• Thinking it reacts only after traffic changes
• Confusing manual scaling with predictive scaling
• Believing it only monitors without scaling

2. Which AWS service feature enables Predictive Scaling?

easy

A. Server health monitoring

B. Manual server adjustment

C. Static server allocation

D. Machine learning forecasting

Solution

1. Step 1: Identify the technology behind Predictive Scaling

   Predictive Scaling uses machine learning to forecast future traffic and adjust capacity.

2. Step 2: Match options to this technology

   Only Machine learning forecasting mentions machine learning forecasting, which is the key feature.

3. Final Answer:

   Machine learning forecasting -> Option D

4. Quick Check:

   Predictive Scaling uses machine learning [OK]

Hint: Look for machine learning in options [OK]

Common Mistakes:

• Choosing manual or static options
• Confusing monitoring with forecasting
• Ignoring the predictive aspect

3. Given a sudden traffic spike predicted at 3 PM, what will Predictive Scaling do?

medium

A. Add servers after 3 PM when traffic spikes

B. Increase server capacity before 3 PM to handle traffic

C. Remove servers before 3 PM to save costs

D. Keep server count constant regardless of traffic

Solution

1. Step 1: Understand timing of scaling actions

   Predictive Scaling acts before traffic changes to prepare resources.

2. Step 2: Analyze options for timing and action

   Only Increase server capacity before 3 PM to handle traffic describes increasing capacity before the predicted spike.

3. Final Answer:

   Increase server capacity before 3 PM to handle traffic -> Option B

4. Quick Check:

   Predictive Scaling acts before spikes [OK]

Hint: Predictive means before event, not after [OK]

Common Mistakes:

• Choosing to add servers after spike
• Thinking it reduces servers before traffic
• Assuming no change in server count

4. You configured Predictive Scaling but notice servers are not added before traffic spikes. What is a likely cause?

medium

A. Predictive Scaling is disabled or misconfigured

B. Traffic never changes in your app

C. Manual scaling overrides predictive scaling

D. Server health checks are failing

Solution

1. Step 1: Identify why predictive scaling might not act

   If servers are not added before spikes, predictive scaling may be off or set up incorrectly.

2. Step 2: Evaluate other options

   Traffic changes and manual scaling do not prevent predictive scaling from working if enabled; health checks affect server health but not scaling triggers.

3. Final Answer:

   Predictive Scaling is disabled or misconfigured -> Option A

4. Quick Check:

   Misconfiguration stops predictive scaling [OK]

Hint: Check if predictive scaling is enabled first [OK]

Common Mistakes:

• Blaming traffic patterns
• Ignoring configuration settings
• Confusing health checks with scaling triggers

5. How does Predictive Scaling help balance cost and performance in AWS?

hard

A. By adding servers only after traffic increases to avoid waste

B. By permanently running maximum servers to handle any traffic

C. By forecasting traffic and adjusting capacity before changes to keep apps fast and save money

D. By disabling scaling during low traffic periods

Solution

1. Step 1: Understand cost and performance goals

   Good scaling keeps apps fast and avoids paying for unused servers.

2. Step 2: Match Predictive Scaling behavior

   It forecasts traffic and adjusts capacity before changes, balancing speed and cost.

3. Step 3: Evaluate options

   Only By forecasting traffic and adjusting capacity before changes to keep apps fast and save money describes forecasting and proactive adjustment to balance cost and performance.

4. Final Answer:

   By forecasting traffic and adjusting capacity before changes to keep apps fast and save money -> Option C

5. Quick Check:

   Forecasting balances cost and speed [OK]

Hint: Forecast first, then scale to save money and stay fast [OK]

Common Mistakes:

• Thinking scaling happens only after traffic rises
• Assuming always max servers run
• Believing scaling disables during low traffic