How to store & read events specific to an Aggregate from Kafka in an ES/CQRS application?

Question

I'm trying to understand how other developers are using Kafka for Eventsourcing. I can't find much help googling. My question is specific to rebuilding the state of an Aggregate reading the Events that only that are specific to that aggregate.

Suppose I have a UserAggregate which stores the following events and lets assume my application has a million users and there are about 10 million events stored in total.

UserCreated(email: String, password: String)
UserPasswordUpdated(password: String)
UserEmailUpdated(email: String)

Since in an eventsourced application we have to rebuild the state of an Aggregate every time a new command occurs for an Aggregate. To rebuild the state I have to re-read the Events that occurred for that Aggregate from Kafka.

• I can have 1 topic per Aggregate type. So one topic that stores events related to all UserAggregates. But what if a User's account was created and 3 years later (probably millions of other User events later) user wants to update his/her email. If the events are this far apart how is this going to effect the query each time I want to rebuild the state of a User ?
• If I created a topic or a partition per UserAggregate the problem I read about are that creating a million topics or partitions will exhaust Zookeeper ?