An organization hosts a retail web application on an Azure Virtual Machine Scale Set (VMSS… — Microsoft Azure Developer (AZ-204)

An organization hosts a retail web application on an Azure Virtual Machine Scale Set (VMSS) with a minimum instance count of $2$ and a maximum instance count of $10$ . You configure a scale-out rule that increases the instance count by $2$ instances when the average CPU Percentage is greater than $80\%$ . You need to configure a scale-in rule to reduce the instance count when demand drops. Which of the following scale-in configurations will prevent autoscale flapping? (Select two.)

A scale-in rule that decreases the instance count by $1$ instance when the average CPU Percentage is less than $50\%$ .Cevap
A scale-in rule that decreases the instance count by $2$ instances when the average CPU Percentage is less than $35\%$ .Cevap
C
A scale-in rule that decreases the instance count by $1$ instance when the average CPU Percentage is less than $55\%$ .
D
A scale-in rule that decreases the instance count by $2$ instances when the average CPU Percentage is less than $45\%$ .

Cevap

The correct scale-in configurations are a decrease of 1 instance when the CPU is less than 50%, and a decrease of 2 instances when the CPU is less than 35%.

To prevent flapping, the average CPU percentage after a scale-in must be less than the scale-out threshold of

80\%

. For a scale-in of 1 instance from 3 to 2, a scale-in threshold of

50\%

ensures the new CPU is at most

(3 \times 50\%) / 2 = 75\% < 80\%

. For a scale-in of 2 instances from 4 to 2, a scale-in threshold of

35\%

ensures the new CPU is at most

(4 \times 35\%) / 2 = 70\% < 80\%

. Both configurations are stable.

Adım Adım Çözüm

Calculate the maximum workload just before the scale-in event triggers for each option.

For a threshold

T_{in}

N

instances, the workload is

N \times T_{in}

This workload is redistributed across the remaining instances after the scale-in occurs.

Determine the post-scale-in average CPU percentage.

For scale-in by

k

instances, the new average CPU is

(N \times T_{in}) / (N - k)

To avoid flapping, this new value must be strictly less than the scale-out threshold of

80\%

Evaluate the worst-case scenario at the minimum instance count where scaling can occur (

N = 3

for

k=1

, and

N = 4

for

k=2

With

k=1

T_{in}

must be

< 53.33\%

. With

k=2

T_{in}

must be

< 40\%

Applying these limits shows that

50\%

(for

k=1

) and

35\%

(for

k=2

) are safe thresholds, while

55\%

and

45\%

trigger immediate scale-out loops.

Anahtar Kavram

Autoscale flapping prevention and metric threshold calculations

Soru

Cevap

Adım Adım Çözüm

Anahtar Kavram

Sık Yapılan Hatalar

Konular

Bu konudan daha fazla soru