http://stackoverflow.com/questions/26751346/how-do-you-set-up-neo4j-to-work-with-google-compute-engine/26752772#26752772

On Tuesday, January 8, 2013 4:21:19 AM UTC+5:30, Peter Neubauer wrote:
>
> A typical Lab-Friday or Neo4j Hackathon activity! Anyone around here 
> got an invite yet? 
>
> /peter 
>
> Cheers, 
>
> /peter neubauer 
>
> G:  neubauer.peter 
> S:  peter.neubauer 
> P:  +46 704 106975 
> L:   http://www.linkedin.com/in/neubauer 
> T:   @peterneubauer 
>
> Neo4j 1.8 GA - 
> http://www.dzone.com/links/neo4j_18_release_fluent_graph_literacy.html 
>
>
> On Mon, Jan 7, 2013 at 11:48 PM, Michael Hunger 
> <[email protected] <javascript:>> wrote: 
> > writes to N1 should be combined in a single tx (e.g. by using a 
> messaging 
> > system to combine/collect the appropriate messages in bulk). 
> > 
> > Neo4j can easily combine 10-30k operations in a single TX. the smaller 
> the 
> > transaction size the more the actual transaction overhead will cost 
> compared 
> > to the operation time. 
> > 
> > On google I would love to see a test of Neo4j on google compute engine, 
> but 
> > haven't gotten my invite yet :) 
> > 
> > Cheers 
> > 
> > Michael 
> > 
> > Am 07.01.2013 um 21:53 schrieb Jamie Olson: 
> > 
> > I don't know much about the efficiency of App Engine Data Stores or how 
> > neo4j queries it's data, but I'd be somewhat concerned about performance 
> > given that write speed.  You say, "the transaction does not affect other 
> > entities in the system", but won't you be unable to make any other 
> changes 
> > to groups involving those entities until the transaction completes? 
> > 
> > So if "N1 -- R1 -- N2" is an edge, you write a row with key date or some 
> > such and ancestors "N1/R1/N2", you could write multiple changes to N1 or 
> to 
> > edge of type R1 originating at N1 as a single transaction.  You shoud 
> get 
> > good performance only if you always make bulk changes in a single 
> > transaction, right?  You'd run into that 1-10 writes/second when making 
> > multiple changes to edges originating from N1 as separate transactions, 
> > right? 
> > 
> > 
> > On Tuesday, August 28, 2012 9:18:58 AM UTC-4, James Thornton wrote: 
> >> 
> >> 
> >> 
> >> On Tuesday, August 28, 2012 3:49:18 AM UTC-5, Koen wrote: 
> >>> 
> >>> Lets call it....    "The BigGraph" 
> >>> 
> >>> ps.. however i am not really a deep technical programmer (i am more 
> from 
> >>> the extend-a-sample kind of programming still having a linear 
> programming 
> >>> brain thinking still in NS diagrams .. for the people who know what 
> that is 
> >>> ........) 
> >>> .. so what level of coding expertise would be required here ? 
> python... 
> >>> java ... c++ ..other ??    ..   and does this interest anyone out 
> there ?? 
> >> 
> >> 
> >> Neo4j is in Java so if you are extending Neo4j, you would want to use 
> >> Java. A key consideration is how App Engine handles transactions. 
> >> 
> >> App Engine stores entities and properties logically together like an 
> >> object, and it supports blob properties up to 1 MB.  All single entity 
> >> writes are transactional for that entity and its properties, and you 
> can 
> >> explicitly enable multi entity ("cross group") transactions on a per 
> write 
> >> basis. All transactions are optimistic. 
> >> 
> >> App Engine parallelizes  writes and has high write scalability, but 
> it's 
> >> limited to 1-10 writes per second per entity or entity group so this is 
> an 
> >> important consideration when modeling nodes and relationships in a 
> >> transactional system like Neo4j. 
> >> 
> >> You could model nodes and relationships as entities so when you write a 
> >> node--relationship-->node transaction, you'll use a "cross group" 
> >> transaction. The two nodes and relationship will be  limited to 1-10 
> writes 
> >> per second, but the transaction does not affect other entities in the 
> >> system. 
> >> 
> >> The other aspect is indexing. App Engine has its own indexing system so 
> >> you would need to implement that as well. Until the new Neo4j indexing 
> >> framework is ready, it may be easier to do this as a Blueprints 
> >> implementation (https://github.com/tinkerpop/blueprints/wiki/). 
> >> 
> >> - James 
> >> 
> > 
> > -- 
> > 
> > 
> > 
> > 
> > -- 
> > 
> > 
>

-- 
You received this message because you are subscribed to the Google Groups 
"Neo4j" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to