[LU-23] Improve performance for traversing large directory with readdir+ Created: 16/Dec/10  Updated: 11/Sep/21  Resolved: 11/Sep/21

Status: Resolved
Project: Lustre
Component/s: None
Affects Version/s: None
Fix Version/s: None

Type: New Feature Priority: Minor
Reporter: nasf (Inactive) Assignee: WC Triage
Resolution: Won't Do Votes: 0
Labels: None

Issue Links:
Related
is related to LU-14139 batched statahead processing Resolved
is related to LU-14380 Make statahead better support Breadth... Open
Sub-Tasks:
Key
 Summary
 Type
 Status
 Assignee
LU-31 Please inspect the HLD for readdir+ Review task Resolved Eric Barton  
Bugzilla ID: 17,845
Rank (Obsolete): 10513

 Description   

Currently, we have statahead to improve the performance of traversing large directory, but it is not enough, especially under high network latency environment, we need reduce RPC count under such case. Both aggregate_stat (IT_5) and readdir+ are candidate for that, maybe need to be combined together.

 Comments   
Comment by Andreas Dilger [ 11/Sep/21 ]

Having batched statahead is similar to readdir+ but better in a number of ways. This can be driven by the client to request attributes for specific FIDs, in a variety of access patterns (e.g. depth-first or breadth-first directory traversal), rather than only sequential readdir.

Generated at Sat Feb 10 01:02:59 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.