CE: Estimate the vCores needed for the pool as follows:
For vCore-based purchasing model: MAX( < Total number of DBs X average vCore utilization per DB > , < Number of concurrently peaking DBs X Peak vCore utilization per DB) A: Estimate the storage space needed for the pool by adding the number of bytes needed for all the databases in the pool.
Reference:
https://docs.microsoft.com/en-us/azure/azure-sql/database/elastic-pool-overview Basic Concept: This question tests Azure monitoring and performance management, including which metric, log destination, or alerting behavior fits the operational requirement.
Why A and C and E is Correct: total size of all the databases matches the expected DP-300 administration action. total size of all the databases is related to operational monitoring or tuning, but it must match the exact signal needed: query history, resource utilization, wait/blocking detail, or automatic remediation. The question is not asking for a general Azure capability; it is asking for the feature that produces this result: Which three metrics should you use to size the elastic pool to meet the demands of your workload? number of concurrently peaking databases * peak CPU utilization per database matches the expected DP-300 administration action.
CPU percentage is important for database compute monitoring, but it is not the Stream Analytics sizing metric when the question is about streaming units. The question is not asking for a general Azure capability; it is asking for the feature that produces this result: Which three metrics should you use to size the elastic pool to meet the demands of your workload? total number of databases * average CPU utilization per database is correct because it is the feature whose normal purpose matches the stated requirement. CPU percentage is important for database compute monitoring, but it is not the Stream Analytics sizing metric when the question is about streaming units. The scenario wording points to that specific behavior: Which three metrics should you use to size the elastic pool to meet the demands of your workload?
Why B is Wrong: Active geo-replication creates readable secondary databases, but failover groups add the managed listener and automatic failover abstraction needed for application continuity at scale. It would produce a different operational signal than the one needed to investigate, alert, or tune the workload in this question.
Why D is Wrong: A session window groups events separated by periods of inactivity, which is useful for user sessions rather than fixed continuous aggregation. It is useful in other troubleshooting paths, but this scenario requires a more specific monitoring or optimization feature.