[Pgpool-general] Status of pgPool2?

[Tatsuo Ishii]

> Tatsuo, pgPool team,
> 
> Is there a page somewhere showing the current status of pgPool2 feature 
> implementation and issues?  I've not been able to find one at 
> pgpool.projects.postgresql.org.

We have such info internally (in Japanese only) but do not open to
public at this moment. This is just due to major code developers are
all in my company and they are all Japanese. Not to keep the status to
be secret of course:-)

Here is a brief summary of current status of the pgpool
development. Any questions are weilcome.

1) pgpool (pgpool-I)

- maintenance release only
- the latest release is 3.4.1
- in stable state and no major known bugs

2) pgpool-II 1.x

- pgpool-II stable releases
- the latest release is 1.2.1
- there are a few known bugs but they will be fixed in 1.3 (about to be released)

3) pgpool-II 2.0

- next generation of pgpool-II and has been actively developed
 (you can check the cvs repository head to know how is the actual
 code)

- we hope to release beta1 on November 1. The official relese will be
  around middle of November

- major enhancements over pgpool-II 1.x include:

  - add online recovery. this enables adding a new DB node and sync it
    without stopping pgpool

  - enhance the performance of replication. now the write performance
    is half of ordinaly PostgreSQL at the worst case regardless the
    number of DB nodes. Read performance will increase by adding more
    DB nodes. So we hope the performace of read/write mix cases
    increases by adding more DB nodes.

  - enhance the reliabilty of replication. 

    - Any SQL statement is now excuted in an explicit transaction
      block. This is needed to avoid data inconsistency among DB nodes
      in some corner cases.

    - in case of failure, decide the victim node to be detached by
     using "decision by majority" rule

  - better administration

	- re-reload the configuration file without restarting pgpool

	- invoke specified commad when failover or failback occured. This
      will make better co-operation with Slony-I, for example

  - enhance parallel mode

	- allow to coexist partitioned tables and replicated tables. This
      enables avoiding unwanted cross-node joins which will decrease
      performance significantly.

   - more optimizations

- remaining issues include:

  - allow extended protocol handling in the parallel mode

  - SSL support

  - automatic query cache invalidation