You are configuring metric-based autoscale rules for an Azure App Service plan that curren… — Microsoft Azure Developer (AZ-204)

You are configuring metric-based autoscale rules for an Azure App Service plan that currently runs on $1$ instance. To prevent metric flapping when scaling between $1$ and $2$ instances, you need to set appropriate scale-out and scale-in thresholds. Which two of the following threshold configurations, based on the Average CPU percentage metric, will prevent flapping?

Scale-out when CPU percentage is greater than $80\%$ ; scale-in when CPU percentage is less than $35\%$ Answer
B
Scale-out when CPU percentage is greater than $80\%$ ; scale-in when CPU percentage is less than $45\%$
Scale-out when CPU percentage is greater than $70\%$ ; scale-in when CPU percentage is less than $30\%$ Answer
D
Scale-out when CPU percentage is greater than $70\%$ ; scale-in when CPU percentage is less than $60\%$

Answer

The correct threshold configurations are a scale-out threshold of

80\%

with a scale-in threshold of

35\%

, and a scale-out threshold of

70\%

with a scale-in threshold of

30\%

The configurations specifying a scale-out threshold of

80\%

with a scale-in threshold of

35\%

, and a scale-out threshold of

70\%

with a scale-in threshold of

30\%

, prevent flapping. This is because scaling from

1

2

instances divides the workload in half. For a scale-out threshold of

80\%

, the CPU load drops to approximately

40\%

, which is safely above the scale-in threshold of

35\%

. For a scale-out threshold of

70\%

, the CPU load drops to approximately

35\%

, which is safely above the scale-in threshold of

30\%

Step-by-Step Solution

Analyze how the CPU load is distributed when scaling out from

1

2

instances.

The average CPU load per instance is halved because the workload is divided equally between

2

instances.

To determine the new CPU level after a scale-out action occurs at or just above the scale-out threshold.

Calculate the new CPU level after scaling out for a scale-out threshold of

80\%

and

70\%

For an

80\%

threshold, the new CPU is

80\% / 2 = 40\%

. For a

70\%

threshold, the new CPU is

70\% / 2 = 35\%

To identify the minimum post-scale-out CPU percentage that will be observed on the instances.

Compare the post-scale-out CPU level to the scale-in threshold for each configuration.

For the configuration with scale-out

80\%

and scale-in

35\%

40\% > 35\%

(no flapping). For scale-out

70\%

and scale-in

30\%

35\% > 30\%

(no flapping). The other configurations have scale-in thresholds (

45\%

and

60\%

) higher than the post-scale-out levels, which triggers immediate scale-in (flapping).

To select the configurations where the scale-in threshold is lower than the new CPU load, avoiding immediate scale-in.

Key Concept

Configuring scale-out and scale-in thresholds with a sufficient margin to prevent metric flapping when resource count changes.

Question

Answer

Step-by-Step Solution

Key Concept

Common Mistakes

Topics

More questions from this topic