O'Reilly logo
  • Ashish Kaushal thinks this is interesting:



Cover of Kafka: The Definitive Guide


How same single architecture can supply redundancy and scalability? Consider single partition0 with million messages which become resource deprived. If one wants to scale it horizontally it must be divided (say by 2) across two machines. Parition0 should have 500K and Partition1 should have 500K messages. If all the messages are copied across two partitions, now we have redundancy but at the same time we have two bottle necked machines as well. chapter: 2, paragraph: Topics and Partitions