Hi Sandeep In some scenarios we badly have to implement updates for our use cases. We end up overwriting a partition/sub partition in hive rather than overwriting the whole table to have least performance impact. You need to intelligently decide on partitions based on your use case for having a least over head while implementing updates using overwrite.
I have scribbled something long back, it can give you some idea on it http://kickstarthadoop.blogspot.in/2011/06/implementing-basic-sql-update-statement.html Regards, Bejoy KS ________________________________ From: Sandeep Reddy P <[email protected]> To: [email protected]; Bejoy Ks <[email protected]> Sent: Thursday, August 9, 2012 9:19 PM Subject: Re: Hive append support Hi Bejoy, Thanks for the link. When you say updates are not supported directly is there any other way we can update data in HDFS/Hive? On Thu, Aug 9, 2012 at 10:30 AM, Bejoy Ks <[email protected]> wrote: Hi Sandeep > > >If you are looking at inserting more data into existing tables that has data, >the yes it is supported in hive from 0.8 (i guess). You can see the INSERT >INTO documentation here >https://cwiki.apache.org/Hive/languagemanual-dml.html#LanguageManualDML-InsertingdataintoHiveTablesfromqueries > > >Updates are not supported by hive directly. > > >Regards, >Bejoy KS > > > >________________________________ > From: Sandeep Reddy P <[email protected]> >To: [email protected] >Cc: [email protected]; [email protected] >Sent: Thursday, August 9, 2012 7:56 PM >Subject: Hive append support > > > >Hi, >Is there any version of Hive for append operations (Insert or Update) via HiveQL? Ifsoplease point me at any HiveQL documentation that explains how the append operations work? > >-- >Thanks, >sandeep > > > > -- Thanks, sandeep
