Data modeling question

[Greg Saylor]

I have a "problem" that I think Cassandra is well-suited to solve but having some trouble on how to setup the primary key.

Essentially there are 5 columns that I want the primary key to span.   This is specifically (and only) used for de-duplication.  For example:

class IDontKnowWtfIAmDoing

  include Cequel::Record

  column :some_id, :bigint

  column :some_type, :text

  column :something, :text

  column :somewhere_id, :bigint

  column :something_type, :text

  column :wtf_dt, :timestamp

  timestamps

end

But I want a primary key that covers [:some_id, :some_type, :something, :somewhere_id, :something_type]

I have this working fine outside of Cequel, but since "IF NOT EXISTS" is going to be added at some point I'd like to build with as much of Cequel as possible.

All the examples show something like:

  key :id, :uuid

But in this case I want:

   key [ .. bunch of keys ]

Is it possible to model Cequel using this?

Obviously putting a separate primary key would defeat the purpose of using it to deduce across these columns.   I suppose I could create a hash of them, but that seems hack-esque.

- Greg

[Greg Saylor]

I think I've figured it out.

Should anyone come across this later, you can define multiple "key"s so its like this:

class IDontKnowWtfIAmDoing

  include Cequel::Record

  key :some_id, :bigint

  key :some_type, :text

  key :something, :text

  key :somewhere_id, :bigint

  key :something_type, :text

  column :wtf_dt, :timestamp

  timestamps

end

Nifty!

- Greg