Proposal for new repository: streamsx.graphdb

[apyasic]

I am proposing to create a project for integration with graph databases.

Graph databases are widely used in a lot of areas: content management and access control, bioinformatics, cyber security, network asset management, public transportation, social applications and many others. Bringing graph databases to Streams will allow better implementation of applications dealing with inter-connected data sets in areas above.

Technically, such an integration will allow to run queries in a real-time for sub-graph pattern matches, manage/persist continuously evolving inter-connected data sets in real-time, apply graph mining algorithms in a real-time, etc.

As a first step I propose an integration with IBM System G Native Graph Store: http://systemg.research.ibm.com/db-nativestore.html

[chanskw]

+1... think it will be useful to add in this toolkit.

Would the initial contribution contains a source and sink operator for reading and writing to System G? What else will be included in the contribution?

[apyasic]

The initial version is about to include native functions for CRUD operations on graphs.

[chanskw]

Only received one vote on this. Any objection to creating this repository?
Propose to close voting by May 29.

[mikespicer]

How can a customer get System G and what support is available for them?

[apyasic]

The first toolkit version will contain System G Native Store capabilities. The Native Store itself is about to be integrated as a shared library into a toolkit, so it would be transparent to a customer. In fact, the toolkit design will allow to swich graph databases if required with no changes in SPL code. Couple of other native candidates for this toolkit are Neo4J and Titan.

[hildrum]

+1

[mikespicer]

+1

[ddebrunner]

+1

[cancilla]

+1

[leongor]

+1

[chanskw]

Who should we assign as initial committers to this project?

[apyasic]

Leonid Gorelik and myself at this point.

[chanskw]

Closing... created streamsx.graphdb repository.

[ddebrunner]

@apyasic I had someone interested in this toolkit yesterday but there's been no activity since its creation, is there likely to be a contribution or should the repository be deleted?

[apyasic]

@ddebrunner the toolkit is work in progress. The first release is about to be published till the end of October.

[ddebrunner]

Any reason the development can't be done in the open?

[apyasic]

The key reason for delay was in finding a way to publish SystemG libraries as a part of the toolkit which appears to be non-trivial from IP perspective. The best compromise we've found is to publish open source (tinkerpop/blueprint based toolkit version first and in parallel continue with a SystemG direction. As we initially started with SystemG some extra efforts are required to port a toolkit functionality to pure tinkerpop/blueprint implementation.

[mikespicer]

Integration with a graph DB has come up from a couple of Streams customers recently. I have also heard that IBM is hoping to standardize on tinkerpop as the query api for graph db implementations. Is there any development happening on this toolkit and could it be done in the open with tinkerpop?

[apyasic]

@mikespicer the development is in progress. The sources would be published soon.

[chanskw]

I am closing this. Please continue discussion about graph DB in its repository.

Thanks!