Re: Tuning Triangle Joins on Hive

Thanks, Gopal! That helps a lot! --Firas On Wed, Aug 6, 2014 at 2:03 PM, Gopal V wrote: > On 7/31/14, 12:28 PM, Firas Abuzaid wrote: > > We're running various "triangle" join queries on Hive 0.9.0, and we're >> wondering if we can get any better performance. Here's the query we're >> running:

Re: Tuning Triangle Joins on Hive

On 7/31/14, 12:28 PM, Firas Abuzaid wrote: We're running various "triangle" join queries on Hive 0.9.0, and we're wondering if we can get any better performance. Here's the query we're running: SELECT count(*) FROM table r1 JOIN table r2 ON (r1.dst = r2.src) JOIN table r3 ON (r2.dst = r3.src AN

Re: Tuning Triangle Joins on Hive

Thanks, that's very helpful! On Sat, Aug 2, 2014 at 12:47 PM, Lefty Leverenz wrote: > How does indexes work in hive? >> > > See the Indexes design doc > in the Hive > wiki, although it hasn't been updated. > > -- Lefty > > > On Sat, Au

Re: Tuning Triangle Joins on Hive

> > How does indexes work in hive? > See the Indexes design doc in the Hive wiki, although it hasn't been updated. -- Lefty On Sat, Aug 2, 2014 at 2:07 AM, chandra Reddy Bogala < chandra.reddy2...@gmail.com> wrote: > How does indexes

Re: Tuning Triangle Joins on Hive

How does indexes work in hive? I thought file formats like ORC have indexes in each block. But not a separate index that can help query performance. Thanks, Chandra On Fri, Aug 1, 2014 at 9:10 AM, Devopam Mittra wrote: > Please try the following approach and let me know if you are not getting >

Re: Tuning Triangle Joins on Hive

Please try the following approach and let me know if you are not getting better performance: 1. Ensure indexes are present on dst , rsc columns in the respective tables. 2. Create a subset first taking r2 and r2 (i.e.: r3.src > r2.src) in a physical table, and then create index on its new src colu

Tuning Triangle Joins on Hive

Hi, We're running various "triangle" join queries on Hive 0.9.0, and we're wondering if we can get any better performance. Here's the query we're running: SELECT count(*) FROM table r1 JOIN table r2 ON (r1.dst = r2.src) JOIN table r3 ON (r2.dst = r3.src AND r3.dst = r1.src) WHERE r1.src < r2.src