It's in the code, called "MapFile".

You can generate them automatically as output from a mapreduce job by using
"MapFileOutputFormat" instead of "SequenceFileOutputFormat".

Make sure you only create a MapFile when the data being written is already
sorted, of course. MapFile works with the assumption that the data is in
sorted order.

On 7/28/06, Benjamin Reed <[EMAIL PROTECTED]> wrote:

I have heard a rumor about the existence of an indexed SequenceFile that
is
basically a normal SequenceFile with an associated small index file with
list
of offsets to a subset of the keys in the SequenceFile. The index is
created
as a sorted SequenceFile is written.

We have need of an IndexedSequenceFile so any pointers would be helpful.

ben


--
Bryan A. P. Pendleton
Ph: (877) geek-1-bp

Reply via email to