Exam DP-203 topic 2 question 36 discussion

B is the correct answer. Automated (job) clusters always use optimized autoscaling. The type of autoscaling performed on all-purpose clusters depends on the workspace configuration. Standard autoscaling is used by all-purpose clusters in workspaces in the Standard pricing tier. Optimized autoscaling is used by all-purpose clusters in the Azure Databricks Premium Plan. https://docs.databricks.com/clusters/cluster-config-best-practices.html

We definitely need "Optimized Autoscaling" (not Standard Autoscaling) which is only part of Premium Plan. Reason: We need to scale down after 3 min underutilization and Standard Autoscaling only allows scaling down after at least 10 minutes. Standard autoscaling: "Scales down only when the cluster is completely idle and it has been underutilized for the last 10 minutes." https://docs.microsoft.com/en-us/azure/databricks/clusters/configure

naughty Microsoft :)! the right action is Cluster Policy but it require the Premium tier. If you're on the Standard tier, you cannot implement this solution until you upgrade to Premium. so the answer is B. Upgrade workspace1 to the Premium pricing tier.

B and C are both required to do the job .But .the FIRST thing to do before creating the AUTO-SCALING CLUSTER POLICY s to migrate or upgrade to premium tier where AUTO-SCALING is supported or enabled.This explains why B is the valid answer . In conclusion C is part of the process ,however not the first thing to do

B is correct

Option B

I've finally found a valid answer: https://learn.microsoft.com/en-us/azure/databricks/administration-guide/clusters/policies

B doesn't minimize the costs. To support autoscaling all-purpose clusters in Azure Databricks, you need to create a cluster policy that specifies the auto-scaling settings. The cluster policy allows you to specify when to add or remove workers based on the workload on the cluster. For this scenario, the cluster policy should be configured to automatically scale down workers when the cluster is underutilized for three minutes. This will help to minimize costs by reducing the number of idle workers. The policy should also be configured to scale to the maximum number of workers quickly to minimize the time it takes to process workloads. Enabling container services for workspace1 (option A) is not necessary for autoscaling all-purpose clusters. Upgrading workspace1 to the Premium pricing tier (option B) may not be necessary and may not be cost-effective depending on your specific requirements. Setting Cluster Mode to High Concurrency (option C) is not related to autoscaling all-purpose clusters.

B is correct

correct

B is the correct answer

Not sure if this is a valid question anymore. This link shows that the standard pricing tier supports optimised autoscaling. https://databricks.com/product/azure-pricing

They need to use Optimized autoscaling for adapting requirements. - Optimized autoscaling is used by all-purpose clusters in the Azure Databricks Premium Plan. - On job clusters, scales down if the cluster is underutilized over the last 40 seconds. - On all-purpose clusters, scales down if the cluster is underutilized over the last 150 seconds. reference: https://docs.microsoft.com/en-us/azure/databricks/clusters/configure

1. Both standard and premium pricing tire support Autopilot Cluster. Autopilot support Autoscaling and Terminate after X minutes of inactivity. 2. 'Cluster Policies' is only supported by premium pricing tire. Control cost by limiting per cluster maximum cost. 3. standard pricing tire is cheaper than premium pricing tire. Base on these 3 items, I don't figure out why it has to upgrade to Premium pricing tier.

Answer B is correct. One has to check on the documentation. There are two autoscaling solutions: standard autoscaling (Standard Tier) and optimized autoscaling (Premimum Tier). Since there is a requirement of downscaling after three minutes of underutilization, only optimized autoscaling can offer such a solution. https://docs.microsoft.com/en-us/azure/databricks/clusters/configure#optimized-autoscaling On all-purpose clusters, scales down if the cluster is underutilized over the last 150 seconds.

Why is the answer not D? Autoscaling is available in the Standard pricing tier. Since "costs" is also a factor in this question, why upgrade to premium?

Is this correct?