Advanced Patterns, Ecosystem & Observability Quiz

Q1. EASY: Which messaging pattern ensures each message is consumed by only one recipient (and not broadcast to all subscribers)? Publish/subscribe (multiple subscribers receive each message)
Work-queue (each message goes to only one consumer)
Fan-out broadcast (deliver every message to all queues)
Request/reply (one-to-one synchronous communication pattern)

Q2. EASY: Which AMQP exchange type routes a message to queues whose binding key exactly matches the message’s routing key? Headers exchange
Topic exchange
Fan-out exchange
Direct exchange

Q3. HARD: In the context of messaging systems, what does exactly-once delivery semantics guarantee? That each message is delivered at least once (no messages are lost, but duplicates are possible)
That each message is processed only once with no duplicates, even in the event of retries or failures
That each message is delivered at most once (no duplicates, but some messages may be lost)
That each published message is received exactly one time by every subscriber

Q4. HARD: Which Kafka feature allows retaining only the latest record for each key in a topic, enabling state reconstruction for event sourcing use cases? Time-based log retention
Idempotent message production
Log compaction
In-memory caching of messages

Q5. EASY: Which messaging protocol is a lightweight publish/subscribe system often used for IoT devices? AMQP
STOMP
HTTP
MQTT

Q6. EASY: Which mechanism is primarily used to encrypt data in transit for messaging systems? SASL (Simple Authentication and Security Layer)
OAuth 2.0
TLS (Transport Layer Security)
ACLs (Access Control Lists)

Q7. MEDIUM: How does RabbitMQ allow multiple tenants or applications to be isolated on a single cluster? Use a fanout exchange for each tenant to segregate messages
Apply per-tenant ACL rules in one shared namespace
Define separate virtual hosts (vhosts) for each tenant on the broker
Run each tenant on a different network port of the broker

Q8. EASY: In Kafka, what happens to messages that have exceeded the configured retention period of a topic? They are moved to a Dead Letter Queue for later processing
They remain available indefinitely
They are archived to long-term storage automatically
They are deleted and no longer accessible after expiration

Q9. MEDIUM: In a Kafka log-compacted topic, how is a "tombstone" used to mark a record as deleted? By writing a message with that key and a null value (a delete marker) to the topic
By using an administrative tool to remove the record from the log
By adding a special header flag to the message indicating deletion
By waiting for the record to be removed by the time-based retention policy

Q10. MEDIUM: In a streaming platform like Kafka, which metric indicates that a consumer is lagging behind the producer in processing messages? Throughput (messages processed per second)
Queue depth (number of pending messages in the queue)
Consumer lag (difference between producer and consumer positions)
p99 latency (99th percentile processing time)

Q11. MEDIUM: What does "p99 latency" refer to in the context of message processing performance? The 99th percentile latency (99% of messages are processed within this time threshold)
The average latency of processing messages (mean latency)
The median latency of processing (50th percentile latency)
The maximum observed latency for message processing

Q12. MEDIUM: Which change is most likely to increase throughput and allow more consumer parallelism for a Kafka topic? Increase the number of partitions for the topic
Increase the topic's replication factor
Add more consumers to read from a single partition in the same group
Use faster disks (SSDs) on brokers to improve IOPS

Q13. HARD: What is the primary benefit of Kafka’s tiered storage feature? It automatically backs up topic data to a secondary cluster for disaster recovery
It offloads older log segments to inexpensive storage, enabling longer retention with less local disk usage
It caches frequently accessed messages in memory for faster retrieval
It keeps all message data only in RAM instead of on disk

Q14. HARD: How can you achieve delayed message delivery in RabbitMQ (since RabbitMQ does not natively support delayed queues)? By installing the RabbitMQ Delayed Message Exchange plugin and publishing messages with a delay parameter
By enabling a built-in queue message delay option in RabbitMQ's settings
By using a fanout exchange with a message TTL (expiration) to delay forwarding
By using a priority queue and treating higher priority as a delay mechanism

Q15. MEDIUM: What happens when you declare a RabbitMQ queue with a maximum priority and publish messages with varying priority values? Messages with lower priority are dropped if a higher priority message arrives
Messages are delivered strictly in the order they were published (FIFO), regardless of priority
Priority affects which route a message takes but not the order in the queue
Higher priority messages will be delivered before lower priority messages

software-architecture