Content-Disposition: inlinehi guys,anyone thought about MapReduce (google’s core-ish distributed query engine) http://labs.google.com/papers/mapreduce.htmland if it would be easy to do and generalize in q? i haven’t readmuch about distributed programming in k/q.are there big clusters using q?ta, jack
map/reduce is not google’s invention. they just rediscover it and brought it to public attention again.
this is an old idea which was implemented in the lisp machines computers.
in kdb this is already implemented to deal with partitioned databases and multi slaves (threads) setups.
have a look in q.k from the definition of .Q.a0 onwards.
btw, i have done something similar as part of a grid infrastructure package for kdb+ about 4 years ago and was working well but i could not find anybody interested to use it in real life.
felix
Felix LUNGU <felix.lungu> wrote:> map/reduce is not google’s invention. they just rediscover it and brought it> to public attention again.How about Map/Reduce/Merge,http:? Looksinteresting. (This is ACM, but I was able to download the paper, soperhaps it’s unlocked. Let me know if you can’t.)> felix-Wm</http:></felix.lungu>