Hi Atri, Thank you for your comment. Yes, I totally agree with you. Storage layer provides and will provide various properties to indicate the capability of underlying storage. Also, I don't consider join operation push down yet because distributed join operation is mostly done through repartition join or broadcast join that usual underlying storages do not support.
Best regards, Hyunsik On Wed, Jun 24, 2015 at 3:18 AM, Atri Sharma <[email protected]> wrote: > In my opinion, operation pushdown will have to be done carefully given that > we need to be storage aware and make sure that underlying storage supports > operations. > > Does this also involve join pushdowns? > > On Wed, Jun 24, 2015 at 3:31 PM, Hyunsik Choi <[email protected]> wrote: > >> Hi, >> >> I had a talk at Hadoop Summit North America 2015 in Jun 17. This >> presentation is for TAJO-337 and its sub tasks. Here is my >> presentation materials. I hope that it would be helpful for you to >> understand TAJO-337 and our roadmap. >> >> >> http://www.slideshare.net/hyunsikchoi/efficient-in-situ-processing-of-various-storage-types-on-apache-tajo >> >> Best regards, >> Hyunsik >> > > > > -- > Regards, > > Atri > *l'apprenant*
