Transparent failover and restore?

John Allspaw allspaw at
Sat Dec 18 12:33:08 PST 2004

fwiw, our initial testing of ndb mysql clustering was not all that
impressive.  architecturally, it sounds like a super idea. 
performance wise, it just didn't do all that well.


On Sat, 18 Dec 2004 11:25:17 -0800, Josh Berkus <josh at> wrote:
> John, Larry, etc:
> > I think this is something that does not belong in memcached itself, as
> > it isn't really a feature of the cache, but belongs client side so those
> > who wish to distribute items redunandtly into seperate server side may
> > do so.  This is relatively trivial to implement, imagine the following:
> >
> > Client performs hash to determine server to put key in (just like it
> > does now), client puts key/value pair into cache.  Client then performs
> > some kind of second hash that gurantees a different server is selected
> > and puts key/value pair into another servers cache.
> I can see that this would be relatively easy to implement on the client side;
> I can also see room for a server-side daemon that manages it.  Doing it on
> the client side, while relatively easy to hack, has the disadvantage of
> making the server lists maintainence problem even worse.
> I really think there's room for a server-side process that manages the hash
> key distribution of items.    One client of mine already shot down the use of
> memcached for a project because we don't have this.
> --
> Josh Berkus
> Aglio Database Solutions
> San Francisco

More information about the memcached mailing list