Kafka

Question 1

Kafka Broker

Answer 1

Computer/instance running Kafka process

Question 2

Kafka Topics

Answer 2

• Durable logs of events
• Append only
• Can only seek by offset (no index)
• Immutable
• Written to disk

Question 3

Kafka Replication

Answer 3

• Copes of data are put on other brokers/partitions for fault tolerance
• This is one LEAD partition and N-1 followers

Question 4

Kafka Schema Registry

Answer 4

• Database for topic schemas
• JSON, AVRO, Protobuff
• Schema will be cached in producer/consumer to reduce latency

Question 5

Kafka Partitions

Answer 5

• If key is null, messages written round robin
• Add partition key for order (ex. give customer id so all events go to same partition in order)

Question 6

Kafka Connect

Answer 6

• Pluggable connectors that run outside of a cluster
• Source connectors = producer
• Sink connectors = consumer