Blob or columns

[Abhinav Solan <ab...@gmail.com>]

Hi Everyone,

We have a unique situation at my workplace while storing data.
We are using Cassandra as a write through cache where we keep real time
data in Cassandra for around 10 - 20 days and rest we archive it to another
data store as archived data.
The current data which we are going to store has around 20 columns, of
which 3 would be used in primary key and 2 more would be read by systems
which would query while working on Cassandra, rest of the columns are of no
use, only use of these columns are when these would be required
re-construct the data to be archived in our archive store which would be
accessed by our legacy applications.
The question here is -
Should we store these inconsequential data as blob or JSON in one column or
create separate columns for them, which one should be the preferred way
here ?
We are currently using Cassandra 3.x version.

Thanks,
Abhinav

Re: Blob or columns

[Tyler Hobbs <ty...@datastax.com>]

On Fri, Jun 3, 2016 at 10:43 AM, Abhinav Solan <ab...@gmail.com>
wrote:

> Should we store these inconsequential data as blob or JSON in one column
> or create separate columns for them, which one should be the preferred way
> here ?


A blob will be more compact and require less server and driver resources
for serialization and deserialization.  Since you don't need to update
anything in the blob individually, I recommend going with that.


-- 
Tyler Hobbs
DataStax <http://datastax.com/>