SQL: Cloud Resource Utilization Dashboard

Write one SQL query merging CPU and memory metric tables into a unified utilization view: resource_type, application_id, and rounded average usage, filtered to a month and a >50% average threshold. Requirements Write a single SQL query that merges CPU and memory utilization into one unified view for a cloud resource dashboard. The result must have these columns: resource_type — the literal CPU or Memory indicating which metric the row describes. application_id — the application with recorded metrics. average_usage_percentage — the average usage as a percentage, rounded to two decimal places including trailing zeros (e.g. 35.00). Rules: Include only metrics recorded in February 2024. Include only applications whose average usage for that specific resource type is greater than 50%. Sort ascending by application_id, then ascending by resource_type. Schema (both tables share the same shape): cpu_metrics(application_id INT, dt VARCHAR(19), usage_percentage DECIMAL(5,2)) memory_metrics(application_id INT, dt VARCHAR(19), usage_percentage DECIMAL(5,2)) Notes The shape is two grouped aggregates combined with UNION ALL: one branch over cpu_metrics labeled CPU, one over memory_metrics labeled Memory. Each branch filters to February 2024, groups by application_id, and keeps groups with HAVING AVG(usage_percentage) > 50. Use ROUND(AVG(usage_percentage), 2) for the two-decimal output. dt is a VARCHAR(19) timestamp string, so the month filter is a half-open range on dt >= '2024-02-01' AND dt < '2024-03-01' rather than a numeric comparison. Apply the final ORDER BY application_id, resource_type to the combined result, not inside each branch. Preparation Write the two-branch UNION ALL query and confirm the HAVING filter applies per resource type, not across the union. Test an application that exceeds 50% on CPU but not memory to confirm it appears once, only in the CPU branch.