[
https://issues.apache.org/jira/browse/HDDS-2939?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17026479#comment-17026479
]
Supratim Deka edited comment on HDDS-2939 at 1/30/20 7:32 AM:
--------------------------------------------------------------
Thanks for taking a look [~linyiqun].
1. ls operation performance : this is outside the scope of work planned under
HDDS-2939. current focus is to improve the performance of the base namespace
operations - create file and create directory. we can take up improvement of
directory listing('ls') after this work achieves desired results.
2. storing child id in directory table parent entry : this has not been
considered. we might consider such an optimisation only if we are unable to
meet specific latency requirements in name lookup or directory listing.
3. prefix locking overhead : the detailed design of the prefix lock mechanism
has not yet been done. Currently we always operate under a bucket lock in the
OM - even for Object/Key access and not just for FS operations. So this is not
a priority now. the prefix locking work will be taken up later.
At that time, I assume the lock design needs to be such that overhead is
proportional to the number of directories with "active" access - not
proportional to the total number of directories that exist.
was (Author: sdeka):
Thanks for taking a look [~linyiqun].
1. ls operation performance : this is outside the scope of work planned under
HDDS-2939. current focus is to improve the performance of the base namespace
operations - create file and create directory. we can take up improvement of
directory listing('ls') after this work achieves desired results.
2. storing child id in directory table parent entry : this has not been
considered. we might consider such an optimisation only if we are unable to
meet specific latency requirements in name lookup or directory listing.
3. prefix locking overhead : the detailed design of the prefix lock mechanism
has not yet been done. currently we already operate under a bucket lock in the
OM, so this is not a priority now. the prefix locking work will be taken up
later.
At that time, I assume the lock design should be such that the overhead is
proportional to the number of directories with "active" access - not
proportional to the number of directories in the namespace.
> Ozone FS namespace
> ------------------
>
> Key: HDDS-2939
> URL: https://issues.apache.org/jira/browse/HDDS-2939
> Project: Hadoop Distributed Data Store
> Issue Type: Improvement
> Components: Ozone Manager
> Reporter: Supratim Deka
> Assignee: Supratim Deka
> Priority: Major
> Attachments: Ozone FS Namespace Proposal v1.0.docx
>
>
> Create the structures and metadata layout required to support efficient FS
> namespace operations in Ozone - operations involving folders/directories
> required to support the Hadoop compatible Filesystem interface.
> The details are described in the attached document. The work is divided up
> into sub-tasks as per the task list in the document.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: ozone-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: ozone-issues-h...@hadoop.apache.org
