Sorting in Cassandra - cassandra

Question

I'm newbie to Cassandra and I need to model a table for storing towns.
CREATE TABLE towns(
root text,
name text,
type text,
PRIMARY KEY(root,name)
) WITH CLUSTERING ORDER BY (name ASC);
INSERT INTO towns(root,name,type) VALUES('.','New York','city');
INSERT INTO towns(root,name,type) VALUES('.','London','city');
INSERT INTO towns(root,name,type) VALUES('.','Paris','city');
INSERT INTO towns(root,name,type) VALUES('.','Tokio','city');
Is that the only way to get records sorted by towns' names asc in Cassandra?
I tried modelling the table as follows:
CREATE TABLE towns(
name text,
type text,
PRIMARY KEY(name)
);
but I can never get towns sorted as required.

Answer 1

In this way you can have a working sort but you can't scale and you creates hotspot.
Imagine you have a 20 nodes cluster with a Replication Factor of 3 ... 17 out of 20 nodes won't be used to store your data because you have a fixed partition key (very bad idea).
Your partition key is ".", what happens is that cassandra choose which node of the cluster will host your data by calculating an hash of the partition key and then send the replica to 2 other nodes.
Using Cassandra in that way is a very bad idea. I'd rather perform a client side sorting if I can't find some valid partition keys ... imagine that you have to sort cities by nation you could use the nation as partition key
INSERT INTO towns(country,city,type) VALUES('Italy','Rome','city');
INSERT INTO towns(country,city,type) VALUES('Italy','Florence','city');
INSERT INTO towns(country,city,type) VALUES('Italy','Venice','city');
INSERT INTO towns(country,city,type) VALUES('England','London','city');
INSERT INTO towns(country,city,type) VALUES('England','Liverpool','city');
To know if there is a way to organize data the way you need to know the exact query you have to perform.
HTH,
Carlo