Auto scaling App Service in Azure - Step-by-Step Execution

Process Flow - Auto scaling App Service

Start: App Service Running

↓

Monitor Metrics: CPU, Requests

↓

Check Thresholds

↓

Scale Out

↓

Add Instances

↓

Update App Service Instances

↓

Continue Monitoring

The App Service monitors its performance metrics. If usage is high, it adds instances (scale out). If usage is low, it removes instances (scale in). This cycle repeats to keep the app responsive.

Execution Sample

Azure

Monitor CPU > 70%?
  Yes -> Increase instances by 1
Monitor CPU < 30%?
  Yes -> Decrease instances by 1
Else -> No change

This logic checks CPU usage and adjusts the number of App Service instances accordingly.

Process Table

Step	CPU Usage (%)	Condition Checked	Action Taken	Instances Count
1	65	CPU > 70%? No	No change	2
2	75	CPU > 70%? Yes	Scale Out +1	3
3	80	CPU > 70%? Yes	Scale Out +1	4
4	50	CPU < 30%? No	No change	4
5	25	CPU < 30%? Yes	Scale In -1	3
6	20	CPU < 30%? Yes	Scale In -1	2
7	35	CPU < 30%? No	No change	2
8	72	CPU > 70%? Yes	Scale Out +1	3
9	68	CPU > 70%? No	No change	3
10	28	CPU < 30%? Yes	Scale In -1	2

💡 Execution stops after 10 steps of monitoring and scaling decisions.

Status Tracker

Variable	Start	After 1	After 2	After 3	After 4	After 5	After 6	After 7	After 8	After 9	After 10
CPU Usage (%)	65	75	80	50	25	20	35	72	68	28	N/A
Instances Count	2	2	3	4	4	3	2	2	3	3	2

Key Moments - 3 Insights

Why does the number of instances not change when CPU usage is 65%?

Because the condition to scale out requires CPU usage to be greater than 70%, which is false at 65% (see step 1 in execution_table). So no scaling action is taken.

Why do we scale in when CPU usage is 25% but not when it is 35%?

Scaling in happens only if CPU usage is less than 30%. At 25% (step 5), this is true, so instances decrease. At 35% (step 7), the condition is false, so no change occurs.

Can the instances count go below 1 during scaling in?

No, best practice is to keep at least one instance running to avoid downtime. The example keeps instances at minimum 2, never scaling below that.

Visual Quiz - 3 Questions

Look at the execution_table at step 3. What is the instances count after scaling?

A 2

B 3

C 4

D 5

💡 Hint

Check the 'Instances Count' column at step 3 in execution_table.

At which step does the CPU usage first trigger a scale in action?

A Step 5

B Step 4

C Step 6

D Step 7

💡 Hint

Look for the first 'Scale In' action in the 'Action Taken' column in execution_table.

If the CPU threshold for scaling out changed from 70% to 60%, what would happen at step 1?

A No change, instances remain 2

B Scale Out, instances increase to 3

C Scale In, instances decrease to 1

D Error in scaling

💡 Hint

Compare CPU usage at step 1 with new threshold and check scaling condition in execution_table.

Concept Snapshot

Auto scaling App Service:
- Monitors app metrics like CPU usage.
- Scales out (adds instances) if usage > threshold.
- Scales in (removes instances) if usage < threshold.
- Keeps minimum instances to avoid downtime.
- Continuously adjusts to keep app responsive.

Full Transcript

Auto scaling in Azure App Service means the system watches how busy your app is, like checking CPU usage. If the app gets busy (CPU over 70%), it adds more instances to handle the load. If the app is not busy (CPU under 30%), it removes instances to save resources. This process repeats continuously to keep your app running smoothly without manual intervention. The example shows CPU usage changing over time and how the number of instances adjusts accordingly, never going below a safe minimum.

Practice

1. What is the main purpose of auto scaling in Azure App Service?

easy

A. Automatically adjust app resources based on demand

B. Manually increase server size when needed

C. Backup app data to cloud storage

D. Monitor user login activities

Solution

1. Step 1: Understand auto scaling concept

   Auto scaling means the system changes resources automatically without manual intervention.

2. Step 2: Identify the purpose in Azure App Service

   Azure App Service auto scaling adjusts the number of instances based on app demand to maintain performance and control costs.

3. Final Answer:

   Automatically adjust app resources based on demand -> Option A

4. Quick Check:

   Auto scaling = automatic resource adjustment [OK]

Hint: Auto scaling means automatic resource changes [OK]

Common Mistakes:

• Confusing auto scaling with manual scaling
• Thinking auto scaling backs up data
• Assuming auto scaling monitors user activity

2. Which of the following is the correct way to set a CPU-based auto scale rule in Azure App Service?

easy

A. Set a rule to scale in when CPU usage < 90%

B. Set a rule to scale out when memory usage < 30%

C. Set a rule to scale out when CPU usage > 70%

D. Set a rule to scale out when disk space > 80%

Solution

1. Step 1: Identify correct metric for scaling out

   CPU usage is a common metric to trigger scaling out when it exceeds a threshold.

2. Step 2: Check the threshold and direction

   Scaling out happens when CPU usage is high (e.g., > 70%), not low or unrelated metrics like disk space.

3. Final Answer:

   Set a rule to scale out when CPU usage > 70% -> Option C

4. Quick Check:

   Scale out triggers on high CPU [OK]

Hint: Scale out when CPU usage is high, not low [OK]

Common Mistakes:

• Using memory or disk space incorrectly for CPU scaling
• Scaling in when CPU is still high
• Setting wrong thresholds for scaling

3. Given this auto scale setting: minimum instances = 2, maximum instances = 5, scale out rule triggers at CPU > 60%, scale in rule triggers at CPU < 30%. If CPU usage rises from 50% to 65%, what happens?

medium

A. Instances reset to minimum count immediately

B. The number of instances increases by one, up to 5

C. No change in instances because CPU is below 70%

D. The number of instances decreases by one, down to 2

Solution

1. Step 1: Analyze CPU usage against scale out rule

   CPU rose from 50% to 65%, which is above the 60% threshold for scaling out.

2. Step 2: Apply scaling limits

   Instances will increase by one but will not exceed the maximum of 5 instances.

3. Final Answer:

   The number of instances increases by one, up to 5 -> Option B

4. Quick Check:

   CPU > 60% triggers scale out [OK]

Hint: Scale out triggers when CPU crosses threshold upwards [OK]

Common Mistakes:

• Thinking CPU must be above 70% to scale out
• Confusing scale in and scale out rules
• Assuming immediate reset to minimum instances

4. You configured auto scaling with minimum 1 instance and maximum 3 instances. The scale out rule triggers at CPU > 50%. However, your app never scales out even when CPU is 80%. What is the most likely issue?

medium

A. CPU metric is not supported for scaling

B. The minimum instance count is too high

C. Maximum instance count is set too low

D. The scale out rule is disabled or not linked to the App Service

Solution

1. Step 1: Check if scaling rules are active

   If the scale out rule is disabled or not properly linked, scaling won't trigger despite high CPU.

2. Step 2: Verify instance limits and metric support

   Minimum and maximum instance counts are valid, and CPU is a supported metric, so these are unlikely causes.

3. Final Answer:

   The scale out rule is disabled or not linked to the App Service -> Option D

4. Quick Check:

   Active rules needed for scaling [OK]

Hint: Check if scaling rules are enabled and linked [OK]

Common Mistakes:

• Assuming instance counts block scaling
• Believing CPU metric is unsupported
• Ignoring rule activation status

5. You want to optimize costs by setting auto scaling for your Azure App Service. You set minimum instances to 1 and maximum to 10. You add a scale out rule for CPU > 70% and a scale in rule for CPU < 40%. However, during peak hours, your app scales out to 10 instances quickly but never scales in after peak hours. What is the best way to fix this?

hard

A. Add a cooldown period to delay scaling in after peak hours

B. Increase the minimum instance count to 5

C. Remove the scale out rule to prevent scaling beyond 1 instance

D. Set scale in rule to CPU < 80% to scale in faster

Solution

1. Step 1: Understand why scaling in doesn't happen

   Without a cooldown period, rapid fluctuations can prevent scale in or cause instability.

2. Step 2: Apply cooldown period to stabilize scaling

   Adding a cooldown period delays scale in, allowing CPU to stay below threshold steadily before reducing instances.

3. Final Answer:

   Add a cooldown period to delay scaling in after peak hours -> Option A

4. Quick Check:

   Cooldown stabilizes scale in/out actions [OK]

Hint: Use cooldown to prevent rapid scale in/out flips [OK]

Common Mistakes:

• Raising minimum instances unnecessarily
• Removing scale out rule stops scaling out
• Setting scale in threshold too high causes no scale in