[ 
https://issues.apache.org/jira/browse/HDDS-2939?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17149763#comment-17149763
 ] 

maobaolong commented on HDDS-2939:
----------------------------------

[~linyiqun] Thank you for your thought and reply about this Alluxio approach.

{quote}
does it really store all metadata in its master service? 
{quote}
Alluxio master service store its metadata as tired metadata, the top layer is 
CachingInodeStore, the underlying layer is RocksInodeStore, so Alluxio master 
can store all its metadata into RocksInodeStore, never evict its metadata out 
of RocksInodeStore, but for CachingInodeStore, it maintain a hot working set 
here, and a background evict thread will evict the inode when the caching inode 
amount achieve the high watermark until below the low watermark.
{quote}
does it match the case of Ozone Fs?
{quote}
Yeah, it think we can reference it,  design a tiered metastore to store inode 
for Ozone. There are 3 cache class
- InodeCache
- EdgeCache
- ListingCache
With these 3 cache, Ozone can be faster for some rich machine which running the 
OzoneManager. And there is a configure item to specify the maximum for each 
Cache.

I'd love to explain more if it is not clear. BTW, out of this ticket, for now a 
day Ozone, with this caching tier will help Ozone to speed up the metadata 
access operation.

> Ozone FS namespace
> ------------------
>
>                 Key: HDDS-2939
>                 URL: https://issues.apache.org/jira/browse/HDDS-2939
>             Project: Hadoop Distributed Data Store
>          Issue Type: New Feature
>          Components: Ozone Manager
>            Reporter: Supratim Deka
>            Assignee: Rakesh Radhakrishnan
>            Priority: Major
>              Labels: Triaged
>         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: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to