You could try this. C* doesn't do it all for you, but = it will efficiently get you the right data.

-ml

-- put this in <file> and run using 'c= qlsh -f <file>

DROP KEYSPACE latest;

CREATE K= EYSPACE latest WITH replication =3D {

=C2=A0 =C2=A0 'class= 9;: 'SimpleStrategy',=C2=A0

=C2=A0 =C2=A0 'replicatio= n_factor' : 1

};

USE latest;

CREAT= E TABLE time_series (

=C2=A0 =C2=A0 userid text,

=C2=A0= =C2=A0 pkid text,

=C2=A0 =C2=A0 colname map<text, text>,

=C2=A0 =C2=A0 PRIMARY KEY (userid, pkid)

);

UPDATE time_series SET colname =3D colname= + {'200':'Col-Name-1'} WHERE userid =3D 'XYZ' AND = pkid =3D '1000';

UPDATE time_series=C2=A0SET=C2=A0colname= =3D colname + {'201':'Col-Name-2'}=C2=A0WHERE=C2=A0userid = =3D 'XYZ'=C2=A0AND=C2=A0pkid =3D '1001';

UPDATE time_series=C2=A0SET=C2=A0colname =3D colname + {'202':= 'Col-Name-3'}=C2=A0WHERE=C2=A0userid =3D 'XYZ'=C2=A0AND=C2= =A0pkid =3D '1000';

UPDATE time_series=C2=A0SET=C2=A0coln= ame =3D colname + {'203':'Col-Name-4'}=C2=A0WHERE=C2=A0user= id =3D 'XYZ'=C2=A0AND=C2=A0pkid =3D '1000';

UPDATE time_series=C2=A0SET=C2=A0colname =3D colname + {'204':= 'Col-Name-5'}=C2=A0WHERE=C2=A0userid =3D 'XYZ'=C2=A0AND=C2= =A0pkid =3D '1002';

SELECT * FROM time_ser= ies WHERE userid =3D 'XYZ';

-- returns:

-- userid | pkid | colname

<= div>----------+------+-----------------------------------------------------= ------------

-- =C2=A0 =C2=A0XYZ | 1000 | {'200': 'Co= l-Name-1', '202': 'Col-Name-3', '203': 'Col= -Name-4'}

-- =C2=A0 =C2=A0XYZ | 1001 | =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0= =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2= =A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 {'201': 'Col-Name-2'}

-- =C2=A0 =C2=A0XYZ | 1002 | =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0= =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2= =A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 {'204': 'Col-Name-5'}

-- use an app to pop off the latest key/value from the map f= or each row, then sort by key desc.

On Tue, Sep 10, 2013 at 9:21 = AM, Ravikumar Govindarajan <ravikumar.govindarajan@gmail.co= m> wrote:

I have been faced with a pr= oblem of grouping composites on the second-part.

Lets sa= y my CF contains this

TimeSeriesCF
=C2=A0 =C2=A0 =C2= =A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0key: =C2= =A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0UserID

=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0composite-col-name: =C2=A0 =C2=A0TimeUUID:PKID

<= /div>
Some sample data

UserID =3D XYZ =C2=A0 = =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0
=C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2= =A0 =C2=A0 =C2=A0 =C2=A0Time:PKID
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0Col-Name1 =3D 200:10= 00
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0Col-Nam= e2 =3D 201:1001
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0Col-Name3 =3D 202:1000
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0Col-Name4 =3D 203:1000
=C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0Col-Name5 =3D 204:1002

Whenever a time-series query is issued, it should retur= n the following in time-desc order.

UserID =3D XYZ=
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 Col-Name5 =3D 2= 04:1002
=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 Col-Name= 4 =3D 203:1000

=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 Col-Name2 =3D 201:100= 1

Is something like this possible in Cassandra? Is= there a different way to design and achieve the same objective?
=
--
Ravi

=C2=A0 =C2=A0