Exam AZ-303 topic 5 question 105 discussion

The answer is correct. Reason: Scale out occur if ANY condition is met while scale in occur ONLY IF ALL condition are met So on Monday morning start with 3 instance and scale-out to 4 instance at night. as CPU is 85% Then Tuesday morning 4 instance persist with NO Scale-in as CPU% is low BUT Memory% is OK; again by tuesday night 5 instance is SCALED-OUT as CPU% is high. By Wednesday, we operate at 5 instance as we can neither scale-in or scale-out (due to hitting max-instance). Same persist even on Weekend as we cannot scale-in. So we hit max instance by wednesday and stay there for perpetuity i.e. Box1-Box2-Box3 is 5-5-5

Crazy answers, that is not the approach to the question, do not complicate yourself. It starts with 3 instances on Monday at 00:00 and asks the number of instances it will use during the periods below the first time, which is the most logical. Because if you are going to automate something so that on the third day it uses the maximum resources, the solution would be to start the application using 5 instances and it would be a crazy question. The correct answer is 3-4-4.

Correct answer: https://docs.microsoft.com/en-us/azure/azure-monitor/autoscale/autoscale-best-practices#considerations-for-scaling-when-multiple-rules-are-configured-in-a-profile Scale-in does not happen if both thersholds are not met

The answer is correct as the application was always scaled out and there was not a single incident where Azure had a chance to scale it in. Hence, it will always run on 5 instances which are the maximum set.

The answer should be 5-5-5. If you started the app1 on Monday 00:00, there will be 3 instances (no rules applied) On Monday 08:00, there will be 4 instances (+1) On Tuesday 00:00, there will be 4 instances (no rules applied, no change) On Tuesday 08:00, there will be 5 instances (+1) ... On Friday 08:00, there will be 5 instances (max instances is 5) On Saturday 00:00, there will be 5 instances (all rules don't applied, no change) ... On next Monday, there will be 5 instances and it will stay infinite. Now the question is "Identify the max number of instances that will be used" and the answer is always 5 for all periods of utilization.

I think Typo in answer ..... it should be 3-3-3 Considerations for scaling when multiple rules are configured in a profile There are cases where you may have to set multiple rules in a profile. The following autoscale rules are used by the autoscale engine when multiple rules are set. On scale-out, autoscale runs if any rule is met. On scale-in, autoscale require all rules to be met. To illustrate, assume that you have the following four autoscale rules: If CPU < 30%, scale-in by 1 If Memory < 50%, scale-in by 1 If CPU > 75%, scale-out by 1 If Memory > 75%, scale-out by 1 Then the follow occurs: If CPU is 76% and Memory is 50%, we scale out. If CPU is 50% and Memory is 76%, we scale out. On the other hand, if CPU is 25% and memory is 51% autoscale does not scale-in. In order to scale-in, CPU must be 29% and Memory 49%.

Hi, this is tricky question including tip: "You expect App1 to be utilized ...". What means App1 max utilization is 100% per all scale set nodes. The utilization metrics are neither related to node nor VM. Otherwise we have to take into consideration 100% utilization per each node what gives us total 500% CPU and memory utilization per 5 nodes. Weird.

Correct answer is 3-4-4

Expected utilization is given for 3 machines in each time frame (assumes that there are only 3 machines working all the time) - in autoscaling case you must remember that presented utilization will be lower if you have scaled out earlier in my opinion, 3-4-3 -> the maximum number of instances that will be used by App1 during the expected periods Mon 00:00 starts with 3 Mon 08:00 scales out from 3 to 4 Tue 00:00 scales in from 4 to 3 (using 4 VM will give lower MEM usage) Tue 08:00 scales out from 3 to 4 ....... Sat 00:00 scales in from 4 to 3 Sat-Sun stays with 3

This answer should be 3/5/3. This is assuming that you are looking at each time slot individually and none relying on the other. On Monday to Friday 00:00 to 07:59 (CPU 25%/ Mem 60%) - scale in is CPU <30% and Mem <50%. This should remain at 3 since it will never scale in as both requirements are not met for scale in and it starts at 3 instances. On Monday to Friday 08:00 to 23:59 (CPU 85%/ Mem 40%) - the CPU usage exceeds the scale-out rule. It will continue to scale out from 3 to 5 since the CPU is over 85% and 5 is the max limit. Scale-out will continue to occur until the max limit is reached or scale-in conditions are met. On Saturday 00:00 to 23:59 (CPU 30% / Mem 55%) - again scale in is CPU <30% and Mem <50%. This should remain at 3 since it will never scale in and it starts at 3 instances.

3 4 4 is my take Sat Sun will have no scale in nor out since the metrics do not meet the rules 0800 - 2359 will have a scale out of 1 instance and then at start of day it will scale in

Example from https://docs.microsoft.com/en-us/azure/azure-monitor/autoscale/autoscale-best-practices#manual-scaling-is-reset-by-autoscale-min-and-max... On scale-out, autoscale runs if any rule is met. On scale-in, autoscale require all rules to be met. To illustrate, assume that you have the following four autoscale rules: If CPU < 30%, scale-in by 1 If Memory < 50%, scale-in by 1 If CPU > 75%, scale-out by 1 If Memory > 75%, scale-out by 1 Then the follow occurs: If CPU is 76% and Memory is 50%, we scale out. If CPU is 50% and Memory is 76%, we scale out. On the other hand, if CPU is 25% and memory is 51% autoscale does not scale-in. In order to scale-in, CPU must be 29% and Memory 49%.

The ask is "You need to identify the "maximum number of instances" that will be used by App1 during the expected periods of utilization." It is 5 5 5. cause max instance is capped at 5. The table I believe is just added as a distraction.

The answer is correct. The autoscale will keep increasing instances every 10 minutes (default interval) until it hits the maximum as long the conditions are met. Because they didn't mention how it will after we get the fourth instance, I suppose that the conditions will be met again and so we got the fifth instance. Scaling in condition will never be met so we will have always 5 instances.

Very ambiguous, what time does the app being implemented? Assume it's Monday? It never said also, we assume the app keeps running all the time after the initial deployment?

Seeing a lot of different answers for this one, so I will give it a shot. A cooldown period is not listed, so we can assume it is default 10 minutes. For scale IN, all scale in conditions have to be met, for scale out, only one condition has to be met. Answers: 3 4 3 (possibly 4) Monday to Friday 00:00-07:59: We start with 3 instances. With that in mind, at an average of 25% CPU and 60% memory utilization across the three instances, we CANNOT scale in because the memory constraint is not met (CPU < 30%, Memory < 50%). Answer is 3. Monday to Friday 08:00-23:59: At an average of 85% CPU and 40% memory utilization across the three instances, we will scale out at the next evaluation cycle since the CPU > 80 constraint is met. Our new CPU utilization will be: (80*3 instances) / now 4 instances = 60%, memory will be: (40*3 instances) / now 4 instances = 30%. Answer is 4. Saturday to Sunday 00:00-23:59: This one is tricky because it's unclear whether we start with the initial instance count of 3, or if the instance count would actually be 4 because of the second part of this question. No scaling occurs.

I think most of you are missing the point of cooldown, if not specified is 10 mins, meaning within an hr more or less, you reach already the maximum and you stay at 5 instances all week long