hive-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Fabio <buso.fa...@gmail.com>
Subject Cascading in Hive metastore
Date Thu, 16 Feb 2017 16:07:24 GMT
Dear Community,

I'm a master thesis student at RISE SICS (Stockholm). For my master 
thesis I'm investigating Hive's internals as we'd like to integrate Hive 
into the Hops platform we have built here. (www.hops.io)
A feature we'd like to have is that when we delete the table data on 
HDFS this also clears Hive's metadata. We can do this as the Hops 
version of HDFS stores metadata on MySQL Cluster 
(https://dev.mysql.com/doc/refman/5.7/en/mysql-cluster.html).

We believe that a nice and clean way to implement this feature is to use 
foreign keys (between HopsFS' metadata db and Hive's metadata db) and 
"on delete cascade" directives. I gave a look to the /package.jdo/ file 
and to the scripts with the DDL that generate the tables in the db and 
you never specify the on delete cascade directive. In the HIVE-3463 
issue you started debating about this topic but then you didn't develop 
further the idea.

My question is, are there any particular reasons why you dropped the idea?

Thanks for the help you'd like to give

--

Fabio Buso


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message