cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From aaron morton <aa...@thelastpickle.com>
Subject Re: Effect of rangequeries with RandomPartitioner
Date Tue, 10 Jul 2012 09:11:03 GMT
Index files map keys (not tokens) to offsets in the data file.

A range scan uses the index file to seek to the start position in the data file and then does
a partial scan of the data file. 

Cheers
-----------------
Aaron Morton
Freelance Developer
@aaronmorton
http://www.thelastpickle.com

On 9/07/2012, at 7:24 PM, prasenjit mukherjee wrote:

> Thanks for the response. Further questions inline..
> 
> On Mon, Jul 9, 2012 at 11:50 AM, samal <samalgorai@gmail.com> wrote:
>>> 1. With RandomPartitioner, on a given node, are the keys  sorted by
>>> their hash_values or original/unhashed keys  ?
>> 
>> hash value,
> 
> 1. Based on the second answer in
> http://stackoverflow.com/questions/2359175/cassandra-file-structure-how-are-the-files-used
> it seems that the index-file ( for a given ssTable ) contains the
> row-key ( and not the hash_keys ).  Or may be I am missing something.
> 
> 2. Do the "keys" in  Index-file ( ref
> http://hi.csdn.net/attachment/201111/28/0_1322461982l3D8.gif )
> actually contain : hash(row_key)+row_key or something like that ?
> Otherwise you need a separate mapping info from hash_bucket -> rows
> for reading.
> 
> -Thanks,
> Prasenjit


Mime
View raw message