Monitor, troubleshoot, and optimize Azure solutions
Free Microsoft Certified: Azure Developer Associate practice — 6 questions on Monitor, troubleshoot, and optimize Azure solutions, with explanations. No sign-up. Full 12-question mixed test →
A web application uses Azure Cache for Redis implementing the cache-aside pattern. During peak traffic, when a popular cache key expires, hundreds of concurrent requests simultaneously miss the cache and hit the backend SQL database at the same instant, causing CPU spikes and timeouts (a cache stampede). Which approach BEST prevents this while keeping the cache-aside pattern?
A distributed lock (using SETNX/SET NX semantics) ensures only one client regenerates the value for an expired key while concurrent callers either wait for a short period or return slightly stale/last-known data, eliminating simultaneous duplicate backend queries. This preserves the client-managed cache-aside pattern while solving the thundering-herd problem.
A .NET application calls a third-party REST API using a standard HttpClient instance. The Application Insights SDK is installed and configured with default settings. Which capability automatically records this outbound call as a dependency (visible in the Application Map) without any additional custom telemetry code?
The Application Insights SDK automatically instruments common dependency calls, including outbound HttpClient requests, SQL calls, and Azure SDK client calls, via its auto-collection dependency module — no custom TrackDependency code is required for standard scenarios.
An Azure Front Door profile caches responses from a backend API. Clients request the same resource path but with varying tracking query string parameters (e.g., ?utm_source=x, ?utm_source=y) that have no effect on the actual response content. The cache hit ratio is very low because Front Door treats each unique query string as a distinct cache entry. Which caching rule configuration fixes this?
Since none of the varying query string parameters affect the response body, setting the caching behavior to 'Ignore Query Strings' makes Front Door cache a single entry per path regardless of query string values, maximizing the hit ratio for this scenario.
You need a KQL query against the Application Insights 'requests' table in Log Analytics that returns the failure rate percentage per operation name over the last 24 hours, sorted from highest to lowest failure rate. Which query is correct?
This query correctly filters to the last 24 hours, uses the boolean 'success' column (comparing to the literal 'false', not the string "false"), computes the failure percentage per operation_Name, rounds it for readability, and sorts descending to surface the worst-performing operations first.
Your team must monitor a critical business workflow that requires: (1) logging into a web app, (2) navigating to a dashboard, and (3) verifying a specific value is rendered — executed on a recurring schedule from multiple global regions. Which Application Insights approach should you configure?
URL ping and Standard availability tests only issue a single HTTP GET/HEAD request and check status codes/response content — they cannot perform multi-step interactions like login and navigation. For complex, stateful workflows, the documented pattern is to write custom code (e.g., using a browser automation library) that executes the steps and reports results using the TrackAvailability() API, typically run on a schedule via a timer-triggered Azure Function from multiple regions.
Your application uses Azure Cache for Redis as a session store where losing session data is unacceptable. The cache is approaching its maxmemory limit. Which maxmemory-policy setting ensures Redis rejects new write commands rather than evicting any existing keys?
The 'noeviction' policy causes Redis to return an error on write commands once maxmemory is reached instead of evicting any keys, which is required when losing existing data (like session state) is unacceptable — the application must handle write failures gracefully instead of risking silent data loss.
Ready for the real thing?
The full course has two full-length practice tests, video lessons for every exam domain, hands-on labs and detailed answer explanations.
Start my full course on Udemy →