>From what I understand each column family has its own file(s).

When hbase lookups a cell in a column family, how does it perform the
lookup? Is there an index on the file, or is it ordered?
What algorithm is used?
Does it load the entire column family file in memory, or scan it?

I know the actual rows are ordered lexically, is it the same for column
family qualifiers?

If someone could also point me to the relevant files that would be great.

Have you looked at any of the source code?

On 6/6/12 10:35 AM, "S Ahmed" <[EMAIL PROTECTED]> wrote:

>From what I understand each column family has its own file(s).
>
>When hbase lookups a cell in a column family, how does it perform the
>lookup? Is there an index on the file, or is it ordered?
>What algorithm is used?
>Does it load the entire column family file in memory, or scan it?
>
>I know the actual rows are ordered lexically, is it the same for column
>family qualifiers?
>
>If someone could also point me to the relevant files that would be great.

On Wed, Jun 6, 2012 at 7:35 AM, S Ahmed <[EMAIL PROTECTED]> wrote:
> When hbase lookups a cell in a column family, how does it perform the
> lookup? Is there an index on the file, or is it ordered?

Ordered.  Has an "index" of sorted keys to offsets.

> What algorithm is used?
> Does it load the entire column family file in memory, or scan it?
>

Scan (file index is kept in LRU).

> I know the actual rows are ordered lexically, is it the same for column
> family qualifiers?
>

Yes.

> If someone could also point me to the relevant files that would be great.

Start with "9.7.5 Store" in
http://hbase.apache.org/book/regions.arch.html  It has pointers into
source.

St.Ack