Technical Lab: Implement Azure Traffic Manager
Questions
Question 1 — Multiple Choice
A global company has endpoints in three regions: East US, West Europe, and Southeast Asia. The requirement is that users should always be directed to the geographically closest endpoint, with automatic fallback if the primary endpoint is unavailable.
Which Azure Traffic Manager routing method meets this requirement?
A) Weighted, with equal weights between endpoints
B) Performance, with active health checks on all endpoints
C) Geographic, with region mapping for each endpoint
D) Priority, with the closest region endpoint set as priority 1
Question 2 — Technical Scenario
An architect configured Traffic Manager with the Weighted method and the following weights:
| Endpoint | Weight |
|---|---|
| app-eastus | 10 |
| app-westeurope | 10 |
| app-brazil | 80 |
During a test, the app-brazil endpoint was deliberately disabled in the portal. The architect expected 100% of traffic to go to the other two endpoints, but noticed that some users in Brazil were still receiving connection errors for several minutes.
What is the most likely cause of this behavior?
A) Traffic Manager doesn't redistribute traffic when an endpoint is manually disabled, only when health checks fail
B) The DNS TTL was still valid in recursive resolvers and clients, keeping the old endpoint IP in cache
C) The Weighted method doesn't support automatic failover between endpoints
D) The app-brazil endpoint continued responding to health checks even after being disabled
Question 3 — True or False
Azure Traffic Manager operates at the DNS layer and therefore doesn't inspect HTTP request content nor directly forward network packets to endpoints. Consequently, it cannot detect failures that occur after DNS resolution, such as endpoint outages that happen between two health check cycles.
Is the statement above true or false?
Question 4 — Technical Scenario
A team needs to perform a progressive rollout of a new API version. The plan is to start with 5% of traffic directed to the new endpoint and gradually increase, keeping the rest on the current endpoint. No geographic or latency criteria should influence routing.
The team configured the following profile:
Routing method: Performance
Endpoint v1 (current): East US region
Endpoint v2 (new): East US 2 region
What's the problem with this configuration?
A) The Performance method doesn't support multiple endpoints in the same geography
B) The Performance method routes based on network latency, not traffic proportion, making percentage control unpredictable
C) Endpoints in the same region cannot coexist in the same Traffic Manager profile
D) The Performance method requires endpoints to be in regions at least 1000 km apart
Question 5 — Multiple Choice
When configuring health checks in a Traffic Manager profile, which statement below correctly describes the service's behavior when facing a degraded endpoint?
A) Traffic Manager immediately removes the endpoint from rotation as soon as the first health check fails
B) Traffic Manager waits for a configurable number of consecutive failures before marking the endpoint as degraded and removing it from rotation
C) Traffic Manager marks the endpoint as degraded after a single timeout, regardless of tolerance settings
D) Traffic Manager continues sending traffic to the degraded endpoint as long as at least one HTTP 200 response is returned in the last 24 hours
Answer Key and Explanations
Answer Key — Question 1
Answer: B
The Performance method directs users to the endpoint with the lowest network latency from their location, using Microsoft's internet latency table. Since it respects endpoint availability via health checks, automatic fallback occurs naturally when an endpoint is unavailable.
The Geographic method (C) maps users to endpoints based on country or continent deterministically, without considering latency and without automatic fallback to other regions by default. The Priority method (D) defines fixed preference order, not proximity. The Weighted method with equal weights (A) would distribute traffic randomly, ignoring geography.
The consequence of using Geographic in this scenario would be that if the endpoint mapped to a region became unavailable, users from that region would receive errors instead of being redirected.
Answer Key — Question 2
Answer: B
Traffic Manager operates entirely at the DNS layer. When an endpoint is disabled or marked as degraded, Traffic Manager starts responding to DNS queries with other endpoints. However, clients that have already resolved the name and stored the result in cache will continue using the old IP until the TTL expires. This behavior is inherent to DNS and independent of the routing method.
Alternative A is false: manual deactivation also stops routing to the endpoint, just like a health check failure. Alternative C is false: the Weighted method supports failover when endpoints are marked as degraded or disabled. Alternative D is false: manually disabled endpoints don't respond to health checks from Traffic Manager's perspective.
The key learning here is that low TTL is a critical design decision in failover strategies with Traffic Manager.
Answer Key — Question 3
True
Traffic Manager is a DNS-based load balancing service. It responds to DNS queries with the address of the endpoint chosen by the configured routing method, but doesn't act as a proxy or traffic gateway. Network traffic goes directly from the client to the endpoint, without passing through Traffic Manager.
Therefore, if an endpoint fails after DNS resolution and before the next health check cycle, Traffic Manager has no way to detect this state immediately. The client will try to connect directly to the endpoint and may receive errors. This is a fundamental architectural limitation that differentiates Traffic Manager from reverse proxy solutions like Azure Application Gateway or Azure Front Door.
Answer Key — Question 4
Answer: B
The Performance method selects the endpoint based on the lowest network latency between the client and available endpoints. Since both endpoints are in eastern US regions with very similar latencies, routing will become unpredictable and won't reflect the desired 5%/95% proportion.
For progressive rollouts with percentage traffic control, the correct method is Weighted, where each endpoint's weight proportionally determines the fraction of traffic received. Alternative A is false: the Performance method has no geographic restriction on endpoint coexistence. Alternative C is false: there's no region restriction in Traffic Manager for endpoint coexistence. Alternative D is false and represents a distractor without technical basis.
Answer Key — Question 5
Answer: B
Traffic Manager uses a configurable fault tolerance logic. The Tolerated number of failures field defines how many consecutive failures are needed before the endpoint is marked as degraded. This prevents momentary oscillations from causing premature endpoint removal from rotation.
Alternative A describes behavior that's too aggressive and would increase system instability. Alternative C ignores the configurability of the mechanism. Alternative D describes non-existent behavior: Traffic Manager doesn't use 24-hour history to keep endpoints in rotation.
Understanding this mechanism is important because probe intervals and failure tolerance directly determine failure detection time (perceived RTO by DNS) and should be dimensioned according to the service SLA.