David Fetter david
Fri Dec 1 08:55:19 PST 2006
On Fri, Dec 01, 2006 at 11:52:04AM -0500, Jan Wieck wrote:
> On 12/1/2006 11:09 AM, David Fetter wrote:
> 
> Statement based replication in an MVCC system is error prone!
> 
> Just consider the following little flow of events:
> 
> X1: begin;
> X1: set transaction isolation level read committed;
> X2: begin;
> X2: set transaction isolation level read committed;
> X1: update foo set bar = 1 where baz = 5;
> X1: commit;
> X2: update foo set bar = 2 where baz < 20;
> X2: commit;
> 
> In a concurrent environment, you have a race condition between X1's
> update and X2's update. How do you know exactly which one happened
> first in the database?
> 
> And now add 10 other transactions updating foo to that.

Alrighty.  The problem at hand is, "how do I get 1024 copies of a
database?"  Do you have some insight into how to do this with Slony-I?

Cheers,
D
-- 
David Fetter <david at fetter.org> http://fetter.org/
phone: +1 415 235 3778        AIM: dfetter666
                              Skype: davidfetter

Remember to vote!



More information about the Slony1-general mailing list