|
First off, I should thank you for your reply. :)
Second - I don't use Apache, just so we're using the same terms.
At one point I was running (on sub-par hardware) multiple webservers
each with their own local mysql server, all running from the same
master. It was hell. Too fragile, etc. I think I've explained it in a
past thread...
With the memcache layer in place, I don't have to bother with adding
mysql servers and hoping that a master can replicate data real-time to
the slaves, or do some funky routing in code to tell it to read from
the local vs. the master... no thanks. Been there, done that.
In this model, the master mysql server would be beefy enough to take
on normal amounts of load generated from the client websites - perhaps
not in peak usage mode - but that is where memcache will pick up all
the slack (and even more.)
That also doesn't help me, because I'd still need redundant masters -
or have one of the slaves begin acting like a master, and all the
slaves switch to that master, etc. I'd like to keep the servers set on
their specific purpose. Beefy boxes for the database, and a
distributed caching mechanism layer prior to that to alleviate any
unnecessary data fetches.
On 10/30/05, ankush grover <ankushgrover1711@xxxxxxxxx> wrote:
> Hey ,
>
> Why don't you do Mysql clustering means on the same server on which apache
> is running you can do mysql clustering.With mysql clustering if one server
> goes down the other is still available and when the server comes back it can
> take the data from the other mysql data node.
|
