I'm trying Cassandra with simple CRUD operations and don't understand how should I model the data.
Let's say, we need to manage simple user data:
UserId | Email | Name
We want to be able to GET information by either UserId or Email. Also we want to be able to change user info, i.e. Email and Name.
That leads me to a dilemma: to query by Email, I should add it to PRIMARY KEY. But if I index it, I won't be able to UPDATE it.
How should I change the data model or indexing to be able to UPDATE the data?
From what I've read, secondary indexes are evil in Cassandra and I shouldn't use them to keep Cassandra's performance on a good level.
Indeed you should not use secondary indexes unless you absolutely have to. But if you need to search by an email, you can create another table with 2 columns - Email and UserId. The primary key will be Email and that is how you will be searching for a UserId by Email. Think of it as of an index in a traditional relational database. Since the Email value should be unique - the lookup table approach should be more efficient than a secondary index.
Once you found UserId by Email you can use it in the queries to the main table.