On 28 Nov 2012, at 14:02, Marco Neumann <marco.neum...@gmail.com> wrote:
> On Wed, Nov 28, 2012 at 8:55 AM, Andy Seaborne <a...@apache.org> wrote: >> On 28/11/12 13:40, Marco Neumann wrote: >>> >>> do you partition the Fuseki instances? >> >> >> No - it's replicas. >> >> Each is complete (so if you loose a machine, the others can answer the same >> queries. You loose the ingress queries on that one machine as it fails). Or >> add more and the load balancers will spread incoming requests. >> >> There's no cross machine query going on. >> > > that would be an awesome development project, we just need some > funding to deliver it right? ;) > We have a current project which is clustering datasets with custom update engine workers which replicate the sparul operations to the other datasets. The clustering is based on jgroups over which we have layered a channel concept which handles replicating via a n-commit concept currently against jetty hosted servlets. This also provides a form of journal to allow replay. > > >> Andy >> >> >>> >>> >>> On Wed, Nov 28, 2012 at 8:34 AM, Andy Seaborne <a...@apache.org> wrote: >>>> >>>> Fuseki instances >>> >>> >>> >>> >> > > > > -- > > > --- > Marco Neumann > KONA