Extending the Stack...?

2 views
Skip to first unread message

Brian Suda

unread,
Nov 29, 2007, 3:47:35 AM11/29/07
to datapor...@googlegroups.com
I was contacted by someone who deals with alot of commiditized
computing, things like Amazon's EC2, where you can create virtual
machines. He asked me why the dataportability.org didn´'t address
things like the OVM (Open Virtual Machine) so you can not only
portably move your data, but the entire environment to another place.

I didn't have a good answer for him, so i said i'd check with the
mailing list. Is there a reason we started and stopped at the points
in the stack that we did? are we focused with just end-user data, or
something else?

thanks,
-brian

--
brian suda
http://suda.co.uk

Paul Jones

unread,
Nov 29, 2007, 3:55:33 AM11/29/07
to datapor...@googlegroups.com
Is the format OVM or OVF? I just did a quick google search to see how wide-spread the format was, and I couldn't actually get any hits on OVM. I did however find an OVF (Open Virtual Machine Format).

OVF by the looks of it is supported by a number of big players in the industry ( http://xml.coverpages.org/ni2007-09-11-a.html), so I'd probably think it would be a good thing to include. The idea of having portable environments sounds pretty cool to me.

Paul.

Danny Ayers

unread,
Nov 29, 2007, 6:15:38 AM11/29/07
to datapor...@googlegroups.com
On 29/11/2007, Brian Suda <brian...@gmail.com> wrote:
>
> I was contacted by someone who deals with alot of commiditized
> computing, things like Amazon's EC2, where you can create virtual
> machines. He asked me why the dataportability.org didn´'t address
> things like the OVM (Open Virtual Machine) so you can not only
> portably move your data, but the entire environment to another place.

It's an interesting angle, but my take would be that when we're
talking about (declarative) data, the processing environment is
orthogonal - and in many cases it's useful to have a separation of
concerns. (ching ching ching goes the buzzword meter :-)

> I didn't have a good answer for him, so i said i'd check with the
> mailing list. Is there a reason we started and stopped at the points
> in the stack that we did? are we focused with just end-user data, or
> something else?

Hmm, my perspective is obviously a bit twisted through being into
Semantic Web tech, but I'd be wary of scoping the stuff as end-user.
Depending on your definition and/or the application, either all data
is end-user, or all data is /not/ end-user. The only difference I can
see here is whether the data is portable and open (i.e. anyone can use
it) or portable and access-controlled (at a personal or organisational
level).

To put that in a real context, in my personal profile (when I get
around to updating it) you will find:

<me> foaf:interest <http://en.wikipedia.org/wiki/Semantic_Web> .

- that latter URI is known to dbpedia, which I can query:

SELECT ?subject ?comment WHERE {
?subject foaf:page <http://en.wikipedia.org/wiki/Semantic_Web> .
?subject rdfs:comment ?comment .
}

giving a result:
http://tinyurl.com/2nrpbd

which is a short description of that topic in lots of languages.

(In principle I could refer to my own profile in the query too, but in
practice the only systems online that allow SPARQL's FROM like this
are a bit too shaky for demos ;-)

In this scenario there's two bits of eminently portable data being
used - my FOAF plus the dbpedia representation of Wikipedia (part of
the Open Linking Data cloud). The latter is published openly, and as
it happens in this case so is the former. But there's no reason this
particular part of my profile couldn't have controls on it requiring
some form of end-user authentication to access (via HTTP Digest or
whatever).

I guess what I'm trying to say is that with the right systems in place
for transparent communication, there's no clear separation between
what we call portable data and any other data - just different kinds
of access control. Standards like OpenID etc can help get those
systems in place.

Cheers,
Danny.

--

http://dannyayers.com

Reply all
Reply to author
Forward
0 new messages