Fuseki vs Rya

[Laura Morales <la...@mail.com>]

Now that Rya has been promoted to top-level project, I'd like to hear your comments about Fuseki vs Rya. Pros&Cons of both, when and why I should use one or the other. Thanks!

Re: Fuseki vs Rya

[James Anderson <an...@gmail.com>]

it occurs to me,

> On 2019-09-28, at 09:37:20, James Anderson <an...@gmail.com> wrote:
> 
> while one might want to compare the two based on the factors
> 
> - capacity
> - capabilities (including conformance)
> - query execution speed
> - statement import rate
> - resource requirements
> 
> it would be difficult to used published reports to compare fuseki and rya.
> the rya performance assessments used lubm, but nothing equivalent is readily found for jena.
> the rya assessment included rdf3x as the foil, but there no comparison between rdf3x and jena is readily found.
> 
> neglecting for the moment issues related to capabilities and import rate, it is possible gain some insight from the comparison between rdf3x and rya which is present in the rya report from 2013 (https://www.usna.edu/Users/cs/adina/research/Rya_ISjournal2013.pdf), on page 25:
>  <page25image624.png>
> 
> the diagram indicates rough parity between rya and rdf3x.
> the report text suggests this explicitly. (p22)
> the text is not explicit as to the respective run-time environment.
> it does report that the rya execution set-up comprised twenty-two total nodes with eight cores each.
> were one to neglect the storage nodes, on the grounds that at the lubm-2000 scale, which was the basis for the comparison, the respective storage requirements were equivalent, the ratio of nodes used to execute a query remains twelve to one.

in addition, that rdf-3x was (and is still?) single-threaded.

> how much that ratio in resources required to achieve performance parity matters will depend on how important capacity is for a given use case.
> 
>> On 2019-09-25, at 06:25:09, Laura Morales <la...@mail.com> wrote:
>> 
>> Now that Rya has been promoted to top-level project, I'd like to hear your comments about Fuseki vs Rya. Pros&Cons of both, when and why I should use one or the other. Thanks!
> 
> 
> 

Re: Fuseki vs Rya

[James Anderson <an...@gmail.com>]

while one might want to compare the two based on the factors

- capacity
- capabilities (including conformance)
- query execution speed
- statement import rate
- resource requirements

it would be difficult to used published reports to compare fuseki and rya.
the rya performance assessments used lubm, but nothing equivalent is readily found for jena.
the rya assessment included rdf3x as the foil, but there no comparison between rdf3x and jena is readily found.

neglecting for the moment issues related to capabilities and import rate, it is possible gain some insight from the comparison between rdf3x and rya which is present in the rya report from 2013 (https://www.usna.edu/Users/cs/adina/research/Rya_ISjournal2013.pdf), on page 25:
 

the diagram indicates rough parity between rya and rdf3x.
the report text suggests this explicitly. (p22)
the text is not explicit as to the respective run-time environment.
it does report that the rya execution set-up comprised twenty-two total nodes with eight cores each.
were one to neglect the storage nodes, on the grounds that at the lubm-2000 scale, which was the basis for the comparison, the respective storage requirements were equivalent, the ratio of nodes used to execute a query remains twelve to one.
how much that ratio in resources required to achieve performance parity matters will depend on how important capacity is for a given use case.

> On 2019-09-25, at 06:25:09, Laura Morales <la...@mail.com> wrote:
> 
> Now that Rya has been promoted to top-level project, I'd like to hear your comments about Fuseki vs Rya. Pros&Cons of both, when and why I should use one or the other. Thanks!

Re: Fuseki vs Rya

[Claude Warren <cl...@xenei.com>]

Based on a single talk at ApacheConNA by Adina Crainiceanu there is not
much difference in functionality, though Rya does not fully implement
SPARQL 1.1, it probably does enough to work for most projects.  Rya does do
some interesting things with data sketches (
http://incubator.apache.org/projects/datasketches.html) like methods to
speed up processing.  It is implemented on Accumulo and MongoDB.

Jena on the other hand is a more mature project and has years of testing
and bug fixing behind it.  It is implemented on several storage layers
(native, SQL, Cassandra) and provides easy extension points to implement
other storage strategies and integrate them into the query engine.

I suspect they both have their place and that it is a matter of what is
most important to your project.  Only you can determine which trade-offs
are important to your project.

Claude

On Wed, Sep 25, 2019 at 5:25 AM Laura Morales <la...@mail.com> wrote:

> Now that Rya has been promoted to top-level project, I'd like to hear your
> comments about Fuseki vs Rya. Pros&Cons of both, when and why I should use
> one or the other. Thanks!
>


-- 
I like: Like Like - The likeliest place on the web
<http://like-like.xenei.com>
LinkedIn: http://www.linkedin.com/in/claudewarren