> > memcachedb is designed for high-frequency writing and reading, when > your data structure is simple but with a high writing load, usually > when your writing load is equal to reading, it works. > So it is good for site counter(that usually slows the whole mysql) and > index servers, also any high writing and reading situation. > > Now memcachedb is working for our high-traffic site countering service > with a writing load of 1500+ per second per daemon, and with a > moderate load of CPU. The response time is less than 50ms when we give > it a apache module httpd frontend. > > Google Account is good example to using Berkeley DB, though without > using memcache front end. Memcachedb has the same architecture goal > with it. > See: > http://www.usenix.org/events/worlds06/tech/prelim_papers/perl/perl.pdf
So does Memcachedb ever write the data back to say a MySQL db, or it by itself is the db?
