You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@hbase.apache.org by DO YUNG YOON <sh...@gmail.com> on 2015/10/07 19:57:03 UTC
Introducing Project S2Graph
Hello Forks.
I want to introduce an open-source distributed graph database s2graph on
HBase.
Here is the github repository and presentations that describe how s2graph
store data as property graph model and production usecases.
github: https://github.com/kakao/s2graph
presentation:
1. hbasecon(2015-05): http://www.slideshare.net/HBaseCon/use-cases-session-5
2. apachecon(2015-09):
http://schd.ws/hosted_files/apachebigdata2015/06/s2graph_apache_con.pdf
It is inspired by "TAO: Facebook`s Distributed Data Store for Social Graph"
but implementation is different and use HBase as storage.
It is linear scalable regarding to both number of server and how many edges
that bfs search traverse on.
It is not complete graph database, but simple layer to transform data into
property graph model and provide fully asynchronous breadth first search on
graph.
currently more than 20 services from my company kakao(
http://www.kakaocorp.com/en/main) are powered by in production and our
average response time stay under 100ms with 20k breadth first search query
per second at peak on more than trillion edges for about 8 months.
if you find it interesting, please leave me comment.
Thanks,
Doyung Yoon
Re: Introducing Project S2Graph
Posted by Nick Dimiduk <nd...@gmail.com>.
Very cool, thanks for posting Doyung Yoon!
On Wed, Oct 7, 2015 at 10:57 AM, DO YUNG YOON <sh...@gmail.com> wrote:
> Hello Forks.
>
> I want to introduce an open-source distributed graph database s2graph on
> HBase.
>
> Here is the github repository and presentations that describe how s2graph
> store data as property graph model and production usecases.
>
> github: https://github.com/kakao/s2graph
> presentation:
> 1. hbasecon(2015-05):
> http://www.slideshare.net/HBaseCon/use-cases-session-5
> 2. apachecon(2015-09):
> http://schd.ws/hosted_files/apachebigdata2015/06/s2graph_apache_con.pdf
>
> It is inspired by "TAO: Facebook`s Distributed Data Store for Social Graph"
> but implementation is different and use HBase as storage.
> It is linear scalable regarding to both number of server and how many edges
> that bfs search traverse on.
> It is not complete graph database, but simple layer to transform data into
> property graph model and provide fully asynchronous breadth first search on
> graph.
>
> currently more than 20 services from my company kakao(
> http://www.kakaocorp.com/en/main) are powered by in production and our
> average response time stay under 100ms with 20k breadth first search query
> per second at peak on more than trillion edges for about 8 months.
>
> if you find it interesting, please leave me comment.
>
> Thanks,
> Doyung Yoon
>
Re: Introducing Project S2Graph
Posted by Stack <st...@duboce.net>.
Thanks for posting Do Yung Yoon.
St.Ack
On Wed, Oct 7, 2015 at 10:57 AM, DO YUNG YOON <sh...@gmail.com> wrote:
> Hello Forks.
>
> I want to introduce an open-source distributed graph database s2graph on
> HBase.
>
> Here is the github repository and presentations that describe how s2graph
> store data as property graph model and production usecases.
>
> github: https://github.com/kakao/s2graph
> presentation:
> 1. hbasecon(2015-05):
> http://www.slideshare.net/HBaseCon/use-cases-session-5
> 2. apachecon(2015-09):
> http://schd.ws/hosted_files/apachebigdata2015/06/s2graph_apache_con.pdf
>
> It is inspired by "TAO: Facebook`s Distributed Data Store for Social Graph"
> but implementation is different and use HBase as storage.
> It is linear scalable regarding to both number of server and how many edges
> that bfs search traverse on.
> It is not complete graph database, but simple layer to transform data into
> property graph model and provide fully asynchronous breadth first search on
> graph.
>
> currently more than 20 services from my company kakao(
> http://www.kakaocorp.com/en/main) are powered by in production and our
> average response time stay under 100ms with 20k breadth first search query
> per second at peak on more than trillion edges for about 8 months.
>
> if you find it interesting, please leave me comment.
>
> Thanks,
> Doyung Yoon
>