[RANGER-4922] Reduce time to find tags associated with multi-level resource

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 3.0.0, 2.6.0
• Component/s: Ranger
• Assignee: Abhay Kulkarni
• Reporter: Abhay Kulkarni
• Created: 29/Aug/24 21:48
• Updated: 23/Oct/24 16:54
• Resolved: 05/Sep/24 22:49

Description

With the following use case:

• Service supports resource hierarchy with more than one level
• Large number of tags are associated with the resources, with majority of tagged resources with values for all levels in resource hierarchy
• Accessed resource does not have values for all levels in the resource hierarchy

the time required to find the tags associated with the accessed resource is significant.

When tested with a large number of tagged Ozone resources (~ 629,000) with approximately 20 tagged volumes and 103 tagged buckets and the rest being keys, the access evaluation times are:

(volume, bucket, key) : requestCount=629118, avgTimeTaken=49911ns
(volume, bucket)      : requestCount=103, avgTimeTaken=10738069ns
(volume)              :
- requestCount=20, avgTimeTaken=21968890ns
- requestCount=1056, avgTimeTaken=13763978ns (repeated requests in previous run multiple times) 

This patch, using filtering and caching technique attempts to reduce this time.