Yes, it is a solution. I just wonder if we use SequenceFile, how to retrieve the original small files? except traversaling the SequenceFile .
2012/6/27 Harsh J <ha...@cloudera.com> > MapFiles were designed for this. It does something similar to what you > have done, keeps an external index of keys. Why do you not wish to use > MapFiles though? > > On Wed, Jun 27, 2012 at 6:48 AM, Lara Dong <lara.d...@gmail.com> wrote: > > As now we use the Sequence File to solve the small file problems, but the > > Sequence File has no index. What I mean is if I transfer my small file > into > > a Sequence File, how do I know where is the position to get my file in > the > > Sequence File? I know the Map File can avoid this problem, but I prefer > to > > use Sequence File. Just now I use Lucene to build an index to record the > > position. Is there a better way? > > > > > > -- > Harsh J >
