Brian, Allow me to share how we are handling this issue. We have a customer that develop and sell a system that control employees in & out. They have many customers with multiple plants & sites. To avoid the related issue, we did a simple workaround, we add a new column at each table. The column name ;-) server_id :-) Now each insert is running like this: INSERT INTO foo SET server_id = @server_id, auto_increment_column = NULL, bla bla bla BinLog format chosen = ROW. The worse and complex scenario is a customer with 17 sites. All of them linked by ADSL Internet ;-) Yes, believe me. The problem of INSERT / DELETE / SELEC was successfuly solved. The only issue to take care is the possibility of race condition on UPDATE ops. Alexandre Em 27/01/2010, às 18:51, Brian Evans escreveu:
I was reading over the replication discussion on maria-developers I thought I'd share some thoughts that are not so technical as I do not understand the back end that well.
Our system is currently a master-master WAN replication setup, each writing to a different DB with an offline process to do copies. Our databases extensively uses auto-increments on tables (more than necessary, but that's another story).
Would future developments allow synchronous replication to occur on a single database when the WAN link has a single point of failure(SPoF)? How might an auto-increment system handle such a failure?
This SPoF is our network link that can be a transient problem out of our control. Both sides may need to keep inserting data during this transient issue. This is usually not an issue 99% of the time, but it's better to prepare for downtime than trying to fix data errors in that 1%.
_______________________________________________ Mailing list: https://launchpad.net/~maria-discuss Post to : maria-discuss@lists.launchpad.net Unsubscribe : https://launchpad.net/~maria-discuss More help : https://help.launchpad.net/ListHelp
