Jackrabbit for digital media content management

[Charles Earl <ch...@gmail.com>]

Hi,
I'm evaluating CMS and versioning systems for the backend of a SaaS
I'm developing for digital media production workflow. Example content
is from media production suites such as Adobe After
Effects/Premier: video, compositions, etc. So some ability to also
track possible binary diffs in binary data (video) as well as metadata
(e.g. modifications to effects applied in a composition, timeline
annotations, etc.).
- Is this appropriate for  jackrabbit? Are there similar efforts built on top of
jackrabbit?
- Do you foresee any performance challenges?
I'd also like the object storage to be in S3/Amazon BlockStore or
similar remote stores.
My first thought was that git (or something based upon it) would do
just as well, java is my comfort zone though.
Charles

Re: Jackrabbit for digital media content management

[Alexander Klimetschek <ak...@day.com>]

On Thu, Jan 29, 2009 at 5:30 PM, Charles Earl <ch...@gmail.com> wrote:
> I'm evaluating CMS and versioning systems for the backend of a SaaS
> I'm developing for digital media production workflow. Example content
> is from media production suites such as Adobe After
> Effects/Premier: video, compositions, etc. So some ability to also
> track possible binary diffs in binary data (video) as well as metadata
> (e.g. modifications to effects applied in a composition, timeline
> annotations, etc.).
> - Is this appropriate for  jackrabbit?

With JCR observation, you can easily handle any changes to metadata or
binary content in the repository. JCR allows you to version your
content easily, albeit things like diffing and merging of (binary)
content is not part of JCR.

> Are there similar efforts built on top of jackrabbit?

Day Software has an enterprise CMS and digital asset management
product on top of Jackrabbit (using the commercial CRX repository that
is based on Jackrabbit) [1].

> - Do you foresee any performance challenges?

Using the file DataStore is really fast, since it basically streams
directly to the disk and since the datastore entries are never
modified, it doesn't have any rewritings. Entries are also stored by
their md5 hash, so if you have your 2GB video file twice in the
repository, the datastore would contain it only once.

> I'd also like the object storage to be in S3/Amazon BlockStore or
> similar remote stores.

Someone already committed an S3 DataStore implementation for
Jackrabbit, see [2] and [3]. DataStore is the persistence abstraction
in Jackrabbit that stores binary streams. I once tried to write a
Persistence Manager that works with S3 [4], but the PM is too
fine-granular for such a high-latency storage implementation (the PM
is the main persistence interface in Jackrabbit that stores all nodes
and properties).

[1] http://www.day.com
[2] http://markmail.org/message/iny24qxsieutfzeq
[3] https://issues.apache.org/jira/browse/JCR-1724
[4] http://issues.apache.org/jira/browse/JCR-1509

Regards,
Alex

-- 
Alexander Klimetschek
alexander.klimetschek@day.com

Re: Jackrabbit for digital media content management

[Charles Earl <ch...@gmail.com>]

On Thu, Jan 29, 2009 at 6:25 PM, Torgeir Veimo <to...@pobox.com> wrote:
>
> On 30 Jan 2009, at 02:30, Charles Earl wrote:
>
>> So some ability to also track possible binary diffs in binary data (video)
>
> What can tracking binary diff be good for? Are you trying to save space by
> storing only deltas?
>
On reflection you're right...the important things to archive are the
original media content and operations that are applied to it.
C
> --
> Torgeir Veimo
> torgeir@pobox.com
>
>
>
>
>

Re: Jackrabbit for digital media content management

[Torgeir Veimo <to...@pobox.com>]

On 30 Jan 2009, at 02:30, Charles Earl wrote:

> So some ability to also track possible binary diffs in binary data  
> (video)

What can tracking binary diff be good for? Are you trying to save  
space by storing only deltas?

-- 
Torgeir Veimo
torgeir@pobox.com