Andrea Cerrito wrote:
>
> I agree with Ken.
> What are the needs for a cdb file when you have a replicated mysql db?
> About speed, if the db is very large, I don't think that a cdb file can be
> faster than a mysql lookup.
>
> About replicated db, a question. I've looked at the code and I think it was
> made for a read-write + read-only system, isn't it? If it is so, why?
>
> I mean, mySQL replication can be applied in an two way read/write system (or
> read/write + n * read-only system).
The mysql folks built it to be a single update server and
multiple update servers. As an after thought they threw
in that you could do multi-way read/write. I'll bet my
last dollar that the single update/multiple read setup
is better tested than two way read/write.
One of the old programming sayings is,
"Keep it simple, Stupid".
I've seen many a project fail when people tried to get
fancy. To me two way read/write is fancy compared
to singup update, multiple read.
On the other hand, projects where they implemented it
the simplest way and held off on all complicated fancy
things, they worked. Then they added new things to the
system as time went on. Turns out, 90% of those fancy
things never get added because they aren't needed.
Simple systems are more stable and easier to maintain.
