Re: Use fuseki2 and virtuoso 7 together.

[Nauman Ramzan <na...@gmail.com>]

Hi Andy,

thank you for the explanation. That is new information for us since you told us that Virtuoso is an old one. We were always researching about triple stores which are promising a high performance and high scalability. Several benchmarktests were always praising Virtuoso on single servers and on distributed systems. Which solution could you offer at the moment which fulfills both requirements performance and scalability? You also said that we could use VirtDataSource and then any backend as a service inside the assembler of Fuseki would this be then in combination with Jenas SDB which isn't supported anymore?

> On May 31, 2015, at 8:57 PM, Andy Seaborne <an...@apache.org> wrote:
> 
>> On 31/05/15 15:52, John A. Fereira wrote:
>> I've used that driver with another semantic web application that has a configurable triple store but not with fuseki directly.  Be aware that you may encounter some sparql syntax difference for inserts/deletes.
>> 
>> -----Original Message-----
>> From: Nauman Ramzan [mailto:nauman.emallates@gmail.com]
>> Sent: Sunday, May 31, 2015 9:40 AM
>> To: users@jena.apache.org
>> Subject: Use fuseki2 and virtuoso 7 together.
>> 
>> Hi all !
>> I am using fuseki 2 and I want to use virtuoso 7 for store triples.
>> Here is Jdbc driver for jena. https://github.com/srdc/virt-jena I want to load this driver in fuseki using fuseki config.ttl file.
>> Thanks
> 
> Hi,
> 
> There are a number of possibilities: using JDBC is just one of them. Fuseki is more setup to handle models and datasets fro the backend provider.
> 
> (it needs you to write some code to use a JDBC connection to a backend virtuoso instance at the moment).
> 
> OpenLink's VirtDataSource or VirtModel together with an assembler to add it as a service to Fuseki might be an easier way to go.  Vistuos may provide that (as their code is GPL2, Jena isn't going to provide linkage). Unfortunately, all the virtuoso looks to be several years old.
> 
> Finally, you can connect direct to virtuoso using SPARQL (no Fuseki required).  Then you will have to be careful about syntax of SPARQL.
> 
> Do let this list know how you get on.
> 
>    Andy

Re: Use fuseki2 and virtuoso 7 together.

[Andy Seaborne <an...@apache.org>]

Hi Nauman,

May the most important question is what are your scalability and 
performance goals? and including hardware goals (=> running costs).

After all, once a system meets those, anything else is unnecessary. 
Different workloads have different characteristics. (A general database 
comment - nothing specific to linked data here).

The newer SPARQL/graph benchmarks, that are more TPC-like, are clearer 
for the workloads that are targeted at.  That said, they also cost 
significant money to get the benchmarks done so that excludes open 
source projects.

Virtuoso can execute SPARQL natively and provides a SPARQL protocol 
endpoint so SDB does not provide anything (and does not scale).

VirtDataSource is a RDF dataset so in theory, you "just" need to plug it 
into Fuseki.  The "just" is that there needs to be Openlink provided 
custom code to wire it into Jena.  Something needs to register Virtuoso 
Query Engine with Jena because Fuseki isn't going to make custom calls 
to their code by naming their classes in Fuseki.  I don't know what 
OpenLink provide and can't check.

Another thing to consider: a Jena application can speak standard SPARQL 
(query, update, graph store protocols) so you can go straight to the 
Virtuoso store.  You don't need Virtuoso specific code to do that.

	Andy


On 01/06/15 05:23, Nauman Ramzan wrote:
> Hi Andy,
>
> thank you for the explanation. That is new information for us since you told us that Virtuoso is an old one. We were always researching about triple stores which are promising a high performance and high scalability. Several benchmarktests were always praising Virtuoso on single servers and on distributed systems. Which solution could you offer at the moment which fulfills both requirements performance and scalability? You also said that we could use VirtDataSource and then any backend as a service inside the assembler of Fuseki would this be then in combination with Jenas SDB which isn't supported anymore?
>
>> On May 31, 2015, at 8:57 PM, Andy Seaborne <an...@apache.org> wrote:
>>
>>> On 31/05/15 15:52, John A. Fereira wrote:
>>> I've used that driver with another semantic web application that has a configurable triple store but not with fuseki directly.  Be aware that you may encounter some sparql syntax difference for inserts/deletes.
>>>
>>> -----Original Message-----
>>> From: Nauman Ramzan [mailto:nauman.emallates@gmail.com]
>>> Sent: Sunday, May 31, 2015 9:40 AM
>>> To: users@jena.apache.org
>>> Subject: Use fuseki2 and virtuoso 7 together.
>>>
>>> Hi all !
>>> I am using fuseki 2 and I want to use virtuoso 7 for store triples.
>>> Here is Jdbc driver for jena. https://github.com/srdc/virt-jena I want to load this driver in fuseki using fuseki config.ttl file.
>>> Thanks
>>
>> Hi,
>>
>> There are a number of possibilities: using JDBC is just one of them. Fuseki is more setup to handle models and datasets fro the backend provider.
>>
>> (it needs you to write some code to use a JDBC connection to a backend virtuoso instance at the moment).
>>
>> OpenLink's VirtDataSource or VirtModel together with an assembler to add it as a service to Fuseki might be an easier way to go.  Vistuos may provide that (as their code is GPL2, Jena isn't going to provide linkage). Unfortunately, all the virtuoso looks to be several years old.
>>
>> Finally, you can connect direct to virtuoso using SPARQL (no Fuseki required).  Then you will have to be careful about syntax of SPARQL.
>>
>> Do let this list know how you get on.
>>
>>     Andy