Postgres-XC, read/write-scalable multi-master symmetric cluster based on PostgreSQL, version 1.0.1 has been released today.
This minor release is based on the latest PostgreSQL 9.1.5+alpha, meaning that all the patches in PostgreSQL 9.1 stable branch have been merged up to commit d10ddf4 (3rd of September 2012).
You can download the source tarball directly from here.
Like PostgreSQL, this tarball contains all the HTML and man documentation.
The documentation of 1.0, including release notes, is available here.
Around 20 bugs have been fixed since 1.0.0, with in particular those fixes:
- Applications like pgadmin had problems to connect to Postgres-XC server.
- Drop of sequence was not managed correctly when its database was dropped
You can find all the details in the release notes here.
The project can be followed on Source Forge:
And a couple of GIT repositories are used for development:
- SourceForge
- Github
- Twitter: @PostgresXCBot, bot giving tweets about the commits in Postgres-XC GIT repository
The project members are currently working hard on the next version of Postgres-XC that will include those features:
- triggers (being implemented)
- Merge with PostgreSQL 9.2 code (already committed)
- RETURNING, WHERE CURRENT OF (being implemented)
- Insure consistency of utilities that cannot run inside transaction block (ex: CREATE DATABASE safely insured in multiple nodes, being implemented)
- Change table distribution type with ALTER TABLE (already committed)
- Support for cursors (already committed)
- Stuff related to node addition and deletion
- and other things…
The project is under the same license as PostgreSQL, and is managed under a single entity called “Postgres-XC Development Group”.
Have fun with this stable release.
Michael
is XC like stado?
https://launchpad.net/stado
Well, not exactly…
And there are many differences, like Postgres-XC being based on the code of PostgreSQL, so you can use the features of PostgreSQL nativaly in Postgres-XC like streaming replication.
You should begin by having a look at that: http://postgres-xc.sourceforge.net/.
Omer,
Stado is for data warehousing reporting type of workloads, not OLTP, which Postgres-XC performs better at. Postgres-XC is also closer to PostgreSQL in terms of functionality coverage.
That said, you can use streaming replication with the individual Stado nodes, just like Postgres-XC. Each underlying data node in Stado is a standard vanilla PostgreSQL instance, it is not a modified version, like Postgres-XC.
I would select the one that is closer to your particular requirements.
thanks
Hello,
Was wondering if the distribution can be based on a single column only – we rely on a PK which is multicolmn – creating surrogate one is an option but not really elegant and would also affect the storage for 10^10-10^11 entries we are to deal with..
Cheers
Yes you can.
XC has a limitation though, you cannot define a distribution for multiple columns yet. But in your case it is OK, so use the keywords DISTRIBUTE BY HASH (a) at the end of CREATE TABLE query:
CREATE TABLE tab (a int) DISTRIBUTE BY HASH (a);
Thanks for the response – so no other option than creating surrogate column as we have –
Create table s (a,b,c,..,
PRIMARY KEY (a, b))
that’s a pity.
Other option would be to hash by ‘b’ which is the voluminous column- would it be possible to do this?
Do you have any plans to support multicolumn or UDF hashing?
Cheers
Yes it is possible on column b.
And yes the core team is planning for multicolumn hashing support.
What is also in the scope is to allow user to define hashing with a custom function, so you could incorporate a udf hash algorithm yourself.
Multicolumn/UDF hash would be perfect – any schedules to do this yet?