On Fri, Jun 27, 2003 at 08:05:02PM -0400, Dave [Hawk-Systems] wrote:
> some searches on this have produced mixed results...
>
> do we have a stable means to replicate transactions between two physical
> servers, preferrably in master - master configuration where updates/inserts can
> be done on either database and the results replicated to the other master.
I've integrated an replication support in my object abstraction layer in php.
(evrything is running over this layer - no direct queries in the application)
in short words:
each table has the fields inode_id (int4) and mtime (timestamp). on each write
operation, the mtime must be updated. from time to time (i.e. once per minute)
an script runs over the db, fetches out all new records and posts them to
the other nodes. this way, evry node can write anytime (masterless). but this
brings some problems on highly-written databases - conflicts may occour.
i didnt find an better solution yet, but still works for all of my apps,
even running over email :)
cu
--
---------------------------------------------------------------------
Enrico Weigelt == metux ITS
Webhosting ab 5 EUR/Monat. UUCP, rawIP und vieles mehr.
phone: +49 36207 519931 www: http://www.metux.de/
fax: +49 36207 519932 email: contact@metux.de
cellphone: +49 174 7066481 smsgate: sms.weigelt@metux.de
---------------------------------------------------------------------
Diese Mail wurde mit UUCP versandt. http://www.metux.de/uucp/