social-mediagoblin
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Social-mediagoblin] Tahoe-LAFS as a document-oriented database


From: Christopher Allan Webber
Subject: Re: [Social-mediagoblin] Tahoe-LAFS as a document-oriented database
Date: Sun, 10 Apr 2011 10:33:35 -0500
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/24.0.50 (gnu/linux)

Rob Myers <address@hidden> writes:

> On 04/10/2011 04:02 PM, Christopher Allan Webber wrote:
>>
>> /media/USER_ID/ENTRY_ID/${RANDOM_UUID}-filename.png
>
> I assume that the filename is irrelevant at the filesystem level, and
> could be mapped to an actual name through the database?
>
> Or am I misunderstanding document based databases?

Good question.  

So the "document database" in this case doesn't store our media files,
it's still used like *sql.  But the representation of the storage path,
as I'm planning it, will be like:

{'filepath': ['directory1', 'directory2', 'filename.jpg']}

That list is "interpreted" by the file store we're using into whatever
it needs to be.  In this case:

/media/directory1/directory2/filename.jpg

In the example described above, with the random UUID, it might end up
like:

{'filepath': ['4d8e5b0048b1520586000000', '4d8ea23c48b15205e2000000',
              'cf8aa462-10ce-4c92-8cbe-96c19857203b-filename.jpg']}

Which would be expanded into:

/media/4d8e5b0048b1520586000000/4d8ea23c48b15205e2000000/cf8aa462-10ce-4c92-8cbe-96c19857203b-filename.jpg

But maybe in tahoe-lafs or eucalyptus it's expanded in a different way.

Am I clearly explaining myself?

... but part of what I'm asking here is about the *main* (not
eucalyptus, not tahoe-lafs) storage system, and whether or not people
will be okay with a filename like

cf8aa462-10ce-4c92-8cbe-96c19857203b-filename.jpg

and if they aren't, are they okay with having a directory for every
file, like

cf8aa462-10ce-4c92-8cbe-96c19857203b/filename.jpg

... which might chew up a lot of inodes, etc.

If I didn't care about people wgetting the files, I'd do the former
thing, with the uuid just tacked onto the filename.  That way I'd always
be safe about totally unique filenames.  But... I mean, *I* wget files a
lot.

There are some important tradeoffs here to consider, but in the long run
we can change our minds on them later.  I'd stil value peoples' input though. 

-- 
𝓒𝓱𝓻𝓲𝓼𝓽𝓸𝓹𝓱𝓮𝓻 𝓐𝓵𝓵𝓪𝓷 𝓦𝓮𝓫𝓫𝓮𝓻



reply via email to

[Prev in Thread] Current Thread [Next in Thread]