Hi all.

I think it is time for us to build consensus on milestone for next release.
Let me suggest few to start this discussion.

1. Basic implementation for Tinkerpop3.

Many people already asked if they can use S2Graph with tinkerpop. I have
played around at https://issues.apache.org/jira/browse/S2GRAPH-72 issue,
but never polished it yet. I am thinking about provide java client that
implement tinkerpop3 interface by call s2graph core, because java has much
wide range of users and it also can be used by scala users too.

Also I have realized that we can take advantage of tinkerpop's
OLAP(Hadoop-Gremlin) ability by providing layer translating GraphIO between
S2Graph's format to Tinkerpop's format.

Also I think existing userbase for gremlin-console, gremlin-server can take
benefits from this and we may hopefully increase our reach to more broad
audience.

2. Various Storage Backend.

We have tested if our interface is pluggable for storage backend by playing
around RocksDBStorage. I think it is time for us actually provide more
storage backend plugins so users who do not use HBase can also be benefit
from S2Graph.

Many startup users who contact me asked if they can use S2Graph with mysql,
postgresql, and redis. Actually, we have implemented RedisStorage
once(S2GRAPH-1), but never merged PR. Since then, our interface has been
changed so we need to make sure this awesome PR included into our codebase.

I think working on storage backend plugin is great way to interact with
code base, so I encourage others(especially committers) to try out this.

I am thinking of followings.
- Any system that JDBC can talk to.
- postgresl
- mysql
- etc...
- Cassandra(for big graph).
- Redis(S2GRAPH-1, for small graph).
- etc...

Personally I think this is most important discussion for now since anyone
interested in S2Graph can see where we are heading and hopefully encourage
more developers to participate in S2Graph.

Please feel free to give any feedback and comment and hope to discuss this
further with all.

Thanks
DOYUNG YOON

Reply via email to