Hbase, multiple tables

[fe2s]

I've got 50 jpa entities that I want to persist to 50 different Hbase tables. It looks like the only option is to create a peristence unit per entity since in Kundera Hbase table is mapped to keyspace. There are several issues:

1. this is very inconvenient. Persistence.xml becomes large and contains 50 almost copy pasted blocks.
2. I have to create 50 EnityManagers where each manager works with its own entity class
3. I guess having 50 EntityManagers implies keeping 50 connection pools to the same Hbase instance. This is a waste of resources and performance problem.

[chhavigangwal]

@fe2s

For issue#1 , you can create one pu and set its keyspace while creating EMF by setting them in external property map :

  Map<String, String> puProperties = new HashMap<String, String>();
  puProperties.put("kundera.keyspace", "KunderaHbaseKeyspace");
  emf = Persistence.createEntityManagerFactory(_PU, puProperties);

For issue#2 and issue#3, I acknowledge it is an issue with Kundera and team will be working on resolving the same.

Thanks,
Chhavi

[gcastigl]

Hi, are there any updates on this? any workaround?

Thanks in advance!

[fnmartinez]

Hi! I'm having this exact same issue. I'd like to be able to have N tables, one per entity, and manage them all with one entity manager. Has there been any progress with these? Is there any workaround I might be missing?

Te most important question would be if every EntityManager spawned would effectively imply a new connection.

Thanks in advance!

[chhavigangwal]

Fixed with Kundera -2.16 release.

[hegdeaditya]

Hi, can anyone please specify how to map different entities to different tables, post fixing of this issue, without creating different persistence units.

[devender-yadav]

@hegdeaditya

You should use kundera-hbase-v2

<dependency>
     <groupId>com.impetus.kundera.client</groupId>
     <artifactId>kundera-hbase-v2</artifactId>
     <version>${kundera.version}</version>
</dependency>

In this, different entity is mapped with different hTable similar to RDBMS (entity mapped with table)

Check wiki for details.

-Dev