[ https://issues.apache.org/jira/browse/RANGER-2203?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16651789#comment-16651789 ]
Pradeep Agrawal commented on RANGER-2203: ----------------------------------------- Steps required before applying the patch: Before upgrading to the code base of this patch(RANGER-2203) please take the backup of your current ranger database first. After taking the backup please delete ‘DB_PATCHES’ and ‘JAVA_PATCHES’ entry from *x_db_version_h* table of ranger db. To delete the entries please login to ranger db and execute the following sql command: * delete from x_db_version_h where version in('DB_PATCHES','JAVA_PATCHES'); > Review and update database schema for ranger policies to minimize database > queries/updates > ------------------------------------------------------------------------------------------ > > Key: RANGER-2203 > URL: https://issues.apache.org/jira/browse/RANGER-2203 > Project: Ranger > Issue Type: Improvement > Components: Ranger > Affects Versions: 0.7.0, master > Reporter: Abhay Kulkarni > Assignee: Abhay Kulkarni > Priority: Major > Fix For: 0.7.2, 2.0.0 > > > Currently, ranger policies are fully normalized and stored in a multiple > Relational database tables. There is a performance overhead incurred when > retrieving a ranger policy, as multiple database accesses are required to > fully reconstruct it. This is significant when there are large ranger > policies (that is, the number of resources addressed by the policy is large), > and/or when there is a large number of ranger policies in an installation. > This Jira tracks alternate design of database schema, where a policy is > stored in a de-normalized way, in its entirely, in one database table > (preferably as a JSON string). -- This message was sent by Atlassian JIRA (v7.6.3#76005)