It's not how Riak does this, and it's not what's going on under the hood with Cassandra, either.
create bucket.users ( id uuid, username text, email text, phone text, primary key (uuid, username) );
That would be exactly how the data was encoded.
What marshray suggested would look like this:
create table bucket.users (name text, field text, value text, primary key ((name, field))); - with a composite partition key. Then you'd have two separate single-cell partitions "jbellis:email" : "jbellis@example.com" and "jbellis:phone" : "555-5555". You don't want to do that in either Riak or Cassandra.
Ah I see what you mean. I thought the intent was to group by some unique ID for the user.
To clarify, this is only the case if you use Cassandra 2.0 transactions; normal batched writes are not atomic in the sense you're probably thinking.
Edit: Actually, they are isolated in the sense of ACID, doesn't matter what order you do to operations, the answer will be the same. But not isolated the way you want isolated described.