hi Pei, Thanks for your interest in the project! Alluxio (fka Tachyon) is a complementary technology, so it would be interesting to see the two projects working well together.
On Fri, Jun 17, 2016 at 10:02 AM, Pei Sun <p...@alluxio.com> wrote: > Dear Arrow developers, > This is Pei from Alluxio Inc. I am interested in knowing more about > Arrow. I have several questions: > 1. If I want to try it out, is there an instruction? There is not, but you can start with the C++ and Java codebases on GitHub. > 2. How is the columnar format achieved in Arrow? After reading this format > <https://github.com/apache/arrow/blob/master/format/Layout.md>, my feeling > is that it is achieved by the way you represent struct in memory. The > format is a little hard to follow. It will be helpful if you can answer > this question before I read it again and again :) Please see slides 11-14 and 19-22 in our presentation from March, and let me know if it is still unclear: http://www.slideshare.net/wesm/apache-arrow-stratahadoop-world-san-jose-2016-60349194 Two main purposes of the columnar layout is both to facilitate CPU cache- and SIMD-friendly analytics as well as to relocate data structures without performing serialization or some other data structure conversion. Thanks, Wes > > > Thanks a lot! Look forward to trying out this interesting project. > > -- > Pei Sun
