Hi Alla, I’m not the expert of the Drill dev team but we have deployed Drill successfully in our production environment. We are a company in the Internet Advertising and Website/App Analyzing field, so we do have a lots of data(100 billions +) need to statistic and report in seconds.
But we don’t use the default parquet file format as it cannot meet some demands. Instead we develop a new file format to serve that, which we call IndexR. It is going to open-source in one or two months. Check here(https://github.com/shunfei/sfmind/blob/master/indexr.about.en.md <https://github.com/shunfei/sfmind/blob/master/indexr.about.en.md>) for a first glance, and tell us if it fits your situation. Regards Flow Wei > On Dec 7, 2016, at 01:06, Alla Gorina <[email protected]> wrote: > > hat Drill performs the best with Parquet file store. And I have two questions:
