In Apache Kafka why can't there be more consumer instances than partitions?

I'm learning about Kafka, reading the introduction section here

https://kafka.apache.org/documentation.html#introduction

specifically the portion about Consumers. In the second to last paragraph in the Introduction it reads

Kafka does it better. By having a notion of parallelism—the partition—within the topics, Kafka is able to provide both ordering guarantees and load balancing over a pool of consumer processes. This is achieved by assigning the partitions in the topic to the consumers in the consumer group so that each partition is consumed by exactly one consumer in the group. By doing this we ensure that the consumer is the only reader of that partition and consumes the data in order. Since there are many partitions this still balances the load over many consumer instances. Note however that there cannot be more consumer instances than partitions.

My confusion stems from that last sentence, because in the image right above that paragraph where the author depicts two consumer groups and a 4-partition topic, there are more consumer instances than partitions!

It also doesn't make sense that there can't be more consumer instances than partitions, because then partitions would be incredibly small and it seems like the overhead in creating a new partition for each consumer instance would bog down Kafka. I understand that partitions are used for fault-tolerance and reducing the load on any one server, but the sentence above does not make sense in the context of a distributed system that's supposed to be able to handle thousands of consumers at a time.

Solution 1:

Ok, to understand it, one needs to understand several parts.

1. In order to provide ordering total order, the message can be sent only to one consumer. Otherwise it would be extremely inefficient, because it would need to wait for all consumers to recieve the message before sending the next one:

However, although the server hands out messages in order, the messages are delivered asynchronously to consumers, so they may arrive out of order on different consumers. This effectively means the ordering of the messages is lost in the presence of parallel consumption. Messaging systems often work around this by having a notion of "exclusive consumer" that allows only one process to consume from a queue, but of course this means that there is no parallelism in processing.

Kafka does it better. By having a notion of parallelism—the partition—within the topics, Kafka is able to provide both ordering guarantees and load balancing over a pool of consumer processes. This is achieved by assigning the partitions in the topic to the consumers in the consumer group so that each partition is consumed by exactly one consumer in the group. By doing this we ensure that the consumer is the only reader of that partition and consumes the data in order. Since there are many partitions this still balances the load over many consumer instances. Note however that there cannot be more consumer instances than partitions.

Kafka only provides a total order over messages within a partition, not between different partitions in a topic.

Also what you think is a performance penalty (multiple partitions) is actually a performance gain, as Kafka can perform actions of different partitions completely in parallel, while waiting for other partitions to finish.

2. The picture show different consumer groups, but the limitation of maximum one consumer per partition is only within a group. You still can have multiple consumer groups.

In the beginning the two scenarios are described:

If all the consumer instances have the same consumer group, then this works just like a traditional queue balancing load over the consumers.

If all the consumer instances have different consumer groups, then this works like publish-subscribe and all messages are broadcast to all consumers.

So, the more subscriber groups you have, the lower the performance is, as kafka needs to replicate the messages to all those groups and guarantee the total order.

On the other hand, the less group, and more partitions you have the more you gain from parallizing the message processing.

Solution 2:

It is important to recall that Kafka keeps one offset per [consumer-group, topic, partition]. That is the reason.

I guess the sentence

Note however that there cannot be more consumer instances than partitions.

is referring to the "automatic consumer group re-balance" mode, the default consumer mode when you just subscribe() some number of consumers to a list of topics.

I assume that because, at least with Kafka 0.9.x, nothing prevents having several consumer instances, members of the same group, reading from the same partition.

You can do something like this in two or more different threads

Properties props = new Properties();
props.put(ConsumerConfig.GROUP_ID_CONFIG, "MyConsumerGroup");
props.put("enable.auto.commit", "false");
consumer = new KafkaConsumer<>(props);
TopicPartition partition0 = new TopicPartition("mytopic", 0);
consumer.assign(Arrays.asList(partition0));
ConsumerRecords<Integer, String> records = consumer.poll(1000);

and you will have two (or more) consumers reading from the same partition.

Now, the "issue" is that both consumers will be sharing the same offset, you don't have other option since there is only one group, topic and partition into play.

If both consumers read the current offset at the same time, then both of them will read the same value, and both of them will get the same messages.

If you want each consumer to read different messages you will have to sync them so only one can fetch and commit the offset at at time.

Solution 3:

Kafka consumer group model is a hybrid of queuing mechanism where the message once read by one consumer instance is deleted from the queue right away and pub/sub mechanism where the message is not deleted until the retention period set or until it expires and is available to all consumer instances until expiry. So, if you have use case where you want to use which is a pub/sub model but want to leverage it as a queuing mechanism you create consumer group for all your consumer instances. Given Kafka distributes partitions among the consumer instances within a single consumer group it is guaranteed that 1 message is only processed once. If Kafka allows you to have more consumer instances within a single consumer group then it beats the purpose of having the consumer group.

Consider this example:

REST API pub1 published 4 messages to topic1 that has 4 partitions part1 thru part4, so each part has 1 message.

You have 2 microservices sub1 and sub2 as subscribers and there are 4 instances of each microservices running.

Now if you create 2 consumer groups, one for each miroservice sub1instance1 will be mapped to part1, sub1instance2 will be mapped to part2, etc Similarly sub2instance1 will be mapped to part1, sub2instance2 mapped to part2 etc.

As long as your consumer instances within each consumer group are less than or equal to the number of partitions, each instance of your microservice is going to process the message only once. In this case sub1instance1 and sub2instance will process msg1 from part1.

If there're more consumer instances than partitions then Kafka will have to assign same partitions to multiple consumer instances so messages will be processed multiple times by each consumer instance mapped to that partition. This is the reason why Kafka prevents us from having more consumer instances within a consumer group than number of partitions within a topic that consumer group is subscribed to.

Hope this makes sense.

Solution 4:

There is a reason why Kafka can not support multiple consumers per partition.

Kafka broker writes data to the file per partition. So let's say if two partitions are configured, broker will create two files and assign multiple consumer groups where messages can be sent.

Now for each partition, only one consumer consumes messages based on the offset of the file. e.g Consumer 1 will first read messages from file offset 0 to 4096. Now these offset are part of the payload so consumer will know which offset to use while requesting for next messages read.

If multiple consumers are reading from same partition then consumer 1 reads from file with offset 0-4096 but consumer 2 will still try to read from offset 0 unless it also receives message sent to consumer 1. Now if same messages are sent to multiple consumers than it is not a load balancing so Kafka has divided them into consumer groups so all consumer groups can receives messages but within consumer group, only one consumer can receive message.