Re: peer-to-peer replication with Postgres

Thanks for the advice.  In that case, I'll stick with the standard
approach of having a single SQL server and several web frontends and
employ a caching mechanism such as memcache as well.  Thanks!

Mike

On Mon, May 10, 2010 at 9:30 AM, Scott Marlowe <scott.marlowe@gmail.com> wrote:
> On Fri, May 7, 2010 at 10:12 PM, Mike Christensen <mike@kitchenpc.com> wrote:
>> I'm considering using a cloud hosting solution for my website.  It
>> will probably be either Amazon, Rackspace or Hosting.com.  I'm still
>> comparing.  Either way, my site will consist of multiple virtual
>> server instances that I can create and destroy as needed.  Each
>> virtual machine instance will be self contained, meaning it'll run the
>> website and its own instance of postgres.  The website will only talk
>> to the local DB instance.  However, if I'm running several machine
>> instances, I want all the databases to keep in sync preferably with as
>> little lag as possible.
>>
>> This is not a master/slave replication issue where there's one big DB
>> that's always up and everything syncs to, this is basically total
>> peer-to-peer replication where any time data is updated on one server,
>> an update command gets sent to all the other servers.  I would also
>> have to address the issue when I provision a new virtual server, I'd
>> have to import the current data into the DB seamlessly.
>>
>> What's the best way to do this?
>
> I think right now you're stuck coding it up yourself.  No small task.
>
>>  Looks like something like pgPool
>> might be what I want, but I haven't looked into it deeply yet.
>> Thanks!!
>
> The only thing that gets close is bucardo.
>