Exam AZ-204 topic 2 question 48 discussion

I think it should B. CPU usage-based autoscaling D. Predictive autoscaling A. Recurrence profile is used to schedule the scaling of resources at specific times or dates, but it does not meet the requirement to scale up when divers are filling out the questionnaire and scale down after they are complete. It only triggers scaling based on a set schedule, not based on actual usage. C. Fixed date profile is used to specify the number of instances at a specific date and time, but it also does not meet the requirement to dynamically scale based on actual usage. It only sets a fixed number of instances and does not adjust based on changing workloads.

I think it should be A and C instead because: Options "B. CPU usage-based autoscaling" and "D. Predictive autoscaling" are not relevant to this scenario, because they are based on resource utilization or predictive analytics rather than a fixed schedule.

First of all we have App Service here, not a Scale Set, so - predictive autoscaling is not available. If there is anyone that would show me where to set it - I would appreciate it and probably change my mind. Since B is obvious and is definitely one of the solutions, the only ones that left are A and C. Recurrence - I guess it is Repeat specific days profile, because I can't see any other where I could define any recurrence. Can't see possibility to set Recurrence profile to - every 15 days as it is per day of week. So the only way to do it would be fixed days - simply specific days in the year on which you have to set number of instances. That's it, so my answer would be B and C.

A and C. I tried auto-scaling a web-app, and both "Fixed date profiles" and "Recurring profiles" are possible options. "CPU usage-based autoscaling" can be configured as part of "Fixed date profiles" and "Recurring profiles". https://learn.microsoft.com/en-us/azure/azure-monitor/autoscale/autoscale-multiprofile?tabs=templates https://learn.microsoft.com/en-us/azure/azure-monitor/autoscale/autoscale-get-started?toc=%2Fazure%2Fapp-service%2Ftoc.json

Long winded but bare with me:

It should only be B. Not recurrence as it scales on a particular date and time or day of the week. Not Predictive as it uses ML to forecast based on historical CPU usage patterns. Not Fixed as well. https://learn.microsoft.com/en-us/azure/azure-monitor/autoscale/autoscale-predictive

To configure autoscaling for the Azure Web Apps to scale up when divers are filling out the questionnaire and scale down after they are complete, you can use the following auto scaling configurations: A. Recurrence profile - You can set up a recurring schedule to scale out the instances during the times when divers are expected to fill out the questionnaire (e.g., every 15 days after a diving job starts) and scale in when the questionnaire completion period ends. B. CPU usage-based autoscaling - You can configure autoscaling based on CPU usage metrics. When the number of divers filling out the questionnaire increases, it might lead to increased CPU usage, triggering the autoscaling to add more instances. Similarly, when the questionnaire completion period ends, CPU usage might decrease, leading to scaling down the instances. So, the correct answers are A (Recurrence profile) and B (CPU usage-based autoscaling).

See: https://learn.microsoft.com/en-us/azure/azure-monitor/autoscale/autoscale-understanding-settings#autoscale-profiles It is not the fixed-date profile because it is relevant for a single date and time. The profile runs only once The Recurrence profile: A recurrence profile is used for a day or set of days of the week. The schema for a recurring profile doesn't include an end date.

Predictive autoscale uses machine learning to help manage and scale Azure Virtual Machine Scale Sets with cyclical workload patterns. It forecasts the overall CPU load to your virtual machine scale set, based on your historical CPU usage patterns. It predicts the overall CPU load by observing and learning from historical usage. This process ensures that scale-out occurs in time to meet the demand. Predictive autoscale needs a minimum of 7 days of history to provide predictions. The most accurate results come from 15 days of historical data.

A and B, For autoscaling in the context of Azure Web Apps with the described requirements, the best options would be: B. CPU usage-based autoscaling This would allow you to scale up the instance count when there is increased load during the questionnaire filling periods. A. Recurrence profile You can schedule the autoscaling based on a recurrence profile to handle the periodic nature of the health questionnaire every 15 days. These two options should provide a suitable configuration for scaling up during questionnaire periods and scaling down afterward.

B. CPU usage-based is a common practice for autoscaling. A. Profile is for date and time-related autoscaling. Do not overthink! Every 15 days => recurring profile Why C & D are wrong? C. Simply do not overthink! D. Predictive autoscaling is for VMSS, not Azure Web Apps.

[15 days after 'each' diving job starts] which means the day to fill the questionnaires can fall in any day, therefore the options that are scaling based on time/dates (A, C) cannot be the solution.

Hope don't have this question in my exam.

https://learn.microsoft.com/en-us/azure/azure-monitor/autoscale/autoscale-predictive Only applies to VM Scale sets

The usage schedule is regular, but we don't know the actual dates on which this happens and if all users fill them out on the same day. But the CPU will go up on those days (A) and the system can work out patterns after it has consumed enough usage data to predict when the spikes will happen (D)

Each driver needs to fill a profile every 15 days after that driver's job starts. Since the company has many drivers, there are no fixed dates or no recurring date patterns. Therefore, options linked to dates are incorrect. We can only use CPU-based or predictive autoscaling.

B. CPU usage-based autoscaling This is one of the most common ways to autoscale an Azure Web App. You can set a threshold for CPU usage, and when the CPU usage goes above that threshold, Azure will automatically add more instances. Similarly, when the CPU usage drops below a certain level, Azure will remove instances. This can be particularly useful if filling out the questionnaires is a resource-intensive operation. D. Predictive autoscaling Predictive autoscaling uses machine learning algorithms to predict future demand based on past usage patterns. If you have historical data that shows spikes in usage around the times when divers are filling out the questionnaire, predictive autoscaling can preemptively scale up before the demand hits and scale down afterward.