Big question. Details matter when considering the amount of work so I
can only make general comments.
What's the interface to the graph ending part? The public blogs (from
2016/08) suggest its stored in a vertex table and an edge table held in
the column store. The "graph algorthms" part is opaque.
Some snooping around says the current version has partial openCypher
support. Mapping to that, and using the property graphs for easy mapping
to/from SPARQL. Map the basic graph pattern to openCypther - or go to
those tables, that's jena-sdb.
Oracle use Jena for their SPARQL - they plugin in at the basic graph
pattern (or at least that was the state when I last heard).
Then the majority of SPARQL is Jena ARQ.
Andy
On 03/01/18 23:02, Dimov, Stefan wrote:
Thanks Andy,
I wanna clarify my questions.
In case we don’t use TDB, the alternative is already chosen (whether we like it
or not). It’s a specific platform, that has both Graph and Relational DB that
are very fast.
They both don’t have SPARQL adapters.
We want to avoid implementing SPARQL layers, so my question is:
In which case I would have less efforts (NOT implementing SPARQL layer),
Relational DB adapter or Graph DB adapter, or in both cases I will need to
implement and maintain the SPARQL layer?
Regards,
Stefan
On 12/22/17, 4:41 PM, "Andy Seaborne" <[email protected]> wrote:
On 23/12/17 00:15, Dimov, Stefan wrote:
> Thanks Andy,
>
> I didn’t quite understand this:
>
> “1/ Use SPARQL (interface RDFConnection) to connect to a remote triple
> store. SPARQL tripe stores are very good at implementing the
standards.”
>
> Does it mean that the graph DB I wanna use, should (out-of-the-box)
support triples and SPARQL or this is the part that I should implement in order to
use it with Jena?
There are quiet a few SPARQL triple stores. Choose and use.
RDFConnection is part of the Jena codebase.
>
> Can you, please, point me, to examples how to:
>
> 1. Implement SQL adapter
See the code for jena-sdb. This component is "maintenance only".
> 2. Implement usage of alternative graph DB
Choose your SPARQL triplestore, download and use!
SPARQL and it's protocols are standards. The tripe store can be any one
that implements the standards. In the SPARQL world, compliance levels
are high.
>
> Happy holidays!
> Stefan
>
>
> On 12/21/17, 1:46 AM, "Andy Seaborne" <[email protected]> wrote:
>
>
>
> On 21/12/17 04:41, Dimov, Stefan wrote:
> > Hi all,
> >
> > Where can I find more detailed info on how can use Jena with:
> >
> >
> > 1. RDBMS
> > * Does it support just any RDBMS?
>
> The supported databases are described in the documentation:
>
> https://jena.apache.org/documentation/sdb/databases_supported.html
>
> Note:
> [[
> Use of SDB for new applications is not recommended.
> This component is "maintenance only".
> TDB is faster, more scalable and better supported than SDB.
> ]]
>
> > * If not, and I want to use it with different (from the
supported) RDMS what should I do?
>
> Do you need it to be an SQL database? (Someone was doing HAMA at one
> time but we didn't hear back from them).
>
> You have to write the SQL adapter - SQL syntax, especially for DDL,
is
> quite variable. There are various examples to start with.
>
> But SDB does not have the scale and speed of a native graph
database.
>
> > 1. Alternative (not TDB) graph DB
> > * Is it possible
>
> Yes.
>
> 1/ Use SPARQL (interface RDFConnection) to connect to a remote
triple
> store. SPARQL tripe stores are very good at implementing the
standards.
>
> 2/ See if the store of your choice offers a Jena adapter.
>
> Andy
>
> > * If not, can it easily be done?
> >
> > Regards,
> > Stefan
> >
>
>
