Some thoughts commented inline below..
On 7 April 2010 05:38, Mike Diehl <mdiehl@stripped> wrote:
> Hi all.
> I'm still doing my reading/research before I configure my replicating db
> servers, but I've got a few quick questions.
> For performance and reliability reasons, I'd like to configure circular
> replication between 3 servers, A, B, and C.
> --Circular replication isn't an architecture that's usually chosen for it's
> performance benefits.
> 1. Will it work? If so, is it stable?
> --In the correct scenario's circular replication works well, however just
> be sure it's the right architecture for your scenario. MySQL replication as
> a technology is both stable and well understood.
> 2. What happens when I want to add/remove a host from the circle?
> --It depends on your application's connection, how do you plan to
> distribute your load across the servers?
> 3. Is there a better way to do it? I do NOT want a solution where I have
> write to one server and read from another. The defeats my reliability
> --What do the schema's look like? How is the data accessed? What's the
> expected transactions per second? What's the proportion of reads / writes?
> How big is the database expected to grow?
> --There are a few case's where circular replication might be a suitable
> solution, it really depends on your scenario.. ..I wouldn't necessarily rule
> out a solution where by you write to one server and read from another, you
> can mitigate concerns you have for reliability by combining replication with
> technologies like DRBD, shared disk clusters, and/or load balanced pools.
> Additionally dependent on your answers to the above, there are other
> concepts you can explore like vertical and/or horizontal partitioning, and
> other technologies like MySQL Cluster.. ..if you need some more detail let
> me know.
> Take care and have fun,
> Mike Diehl.
> MySQL Replication Mailing List
> For list archives: http://lists.mysql.com/replication
> To unsubscribe: