3 ms·
The slowness is probably how far apart your data is spread in the google "cloud" over the years. Map/reduce probably gets a heck of workout, imagine their enti
by ck2 14y ago
The slowness is probably how far apart your data is spread in the google "cloud" over the years.
Map/reduce probably gets a heck of workout, imagine their entire global database size, it must be staggering.
Maybe they can fix it by periodically rewriting the entire dataset for a single user onto a cluster that is much closer together. Once a year?