RE: DBpedia now available as triple pattern fragments

On 31 Okt 2014 at 13:23, Kingsley Idehen wrote:
> On 10/31/14 5:33 AM, Ruben Verborgh wrote:
>>>> The utility of that serves what purpose?  I think Ruben clearly
>>>> described this as a complimentary addition.
>>
>> Exactly_because_  I described them as a complementary addition.
>> It wouldn't make sense to host DBpedia as triple pattern fragments
>> if the DBpedia SPARQL endpoint had high availability.
> 
> That statement doesn't reflect what I mean by being complimentary.
> Basically, in my eyes, that means:
> 
> If OpenLink decides to invest in servers and admin to create the 100%
> illusion (as exemplified by the likes Google etc.) then LD Fragments
> would serve no use?

Depending on the cost it requires to achieve that it might turn out to be
that, yes. But I think that's very *very* unlikely since, as you rightly
point out, these are two vastly different approaches which are difficult to
compare. All I wanted is to collect some data to be able to compare them on
one axis.


> I see LD Fragments as a useful compliment, even if DBpedia's SPARQL
> endpoint had what you would perceive as 100% uptime.
> 
>> After all, the DBpedia SPARQL endpoint solves SPARQL queries much faster
>> when it is available-which is not always the case.
> 
> Which has more to do with what's happening to Virtuoso (from time to
> time), and less to do with the notion of a scalable and high
> availability  SPARQL endpoint.
> 
> You do understand that we offer DBpedia deployment services to the
> world, gratis. We could always make a 100% uptime premium service with
> an SLA (Service Level Agreement).

Right. 


> You continue to conflate misunderstanding of our DBpedia terms of a
> service with a fundamental DBMS science issue.
> 
> Having a SPARQL endpoint, at web-scale, supporting ad-hoc queries is a
> specific DBMS science challenge, for which we've always had an answer.

Great to hear that.


> Anyway, I am going to have a DBpedia report prepared, and posted to the
> appropriate forums.

I'm looking forward to see that report. Please post it to this list as well.


> Can't we all just get along, without distracting in fighting? We are all
> trying to make this Linked Open Data aspect of the World Wide Web a
> reality, right?

No one was fighting.. or intended to, IMO, anyway. We are all aware that LDF
and SPARQL endpoints are very different things, just like apple and oranges
are. Nevertheless I'm sure you agree that it is possible to compare them
(LDF and SPARQL, apple and oranges) on various dimensions.


--
Markus Lanthaler
@markuslanthaler