I know it's deterministic, but it's effectively random because you cannot predict which server a certain item goes to by looking at its key, unless you're the kind of guy that can calculate the hashcode of a string in your head. :-)
/Henrik On Mon, Jan 5, 2009 at 21:59, Jason Rimmer <[email protected]> wrote: > > Given a list of servers a client will place data on a single server > via key-specific algorithmic determination rather than randomly. This > ensures that when presented with the same key a request for retrieval > is made from the same/correct server. This also enables getting and > setting by different clients, such as between java and perl as long as > they use the same algorithm for determination (this also presumes a > shared understanding of the type/structure/whatever of the associated > value). > > See the following FAQ entry for more detail: > http://code.google.com/p/memcached/wiki/FAQ#How_does_memcached_work? > > --- > Jason Rimmer > jrimmer at irth dot net > > On Mon, Jan 5, 2009 at 12:12 PM, Leeladharan MP <[email protected]> > wrote: > <snip> > > On Mon, Jan 5, 2009 at 9:07 PM, Henrik Schröder <[email protected]> > wrote: > >> > >> Hi Leelu, > >> > >> No, you're going up entirely the wrong tree with your solution. In your > >> client, you configure a list of servers that it should use. When you > then > >> use that client, it will distribute items randomly between the servers > you > >> put in the configuration. You can not control which specific server an > item > >> goes to in that case. > </snip> >
