100 million columns vs 100 million rows/keys

[Marcin <ma...@33concept.com>]

Hi guys,

I would like to be able to do slices but have been reading that when 
using too many columns under one key Cassandra becomes very slow? Is it 
true and if so what are the suggestions.

P.S.
The value of the column is just 1 so its quite flat model, reasons 
behind are slicing on name.


cheers,
/Marcin

Re: 100 million columns vs 100 million rows/keys

[Nick Telford <ni...@gmail.com>]

Prior to Cassandra 0.7, there was a limitation of 2GB on row sizes as the
entire row had to fit in memory for compaction. As far as I'm aware, in
Cassandra 0.7, the limit has changed to 2^31 (approximately 2 billion)
columns.

See http://wiki.apache.org/cassandra/CassandraLimitations for more details.

On 26 November 2010 16:45, Marcin <ma...@33concept.com> wrote:

> Hi guys,
>
> I would like to be able to do slices but have been reading that when using
> too many columns under one key Cassandra becomes very slow? Is it true and
> if so what are the suggestions.
>
> P.S.
> The value of the column is just 1 so its quite flat model, reasons behind
> are slicing on name.
>
>
> cheers,
> /Marcin
>