In Hadoop, most files are replicated (space issue?). In MapReduce, there is built in partial memory for restoring (better for recovery of a node failure?). These two optimizations are ways to do computing well even if the hardware isn't that great.
Hi @makingthingsfast, could you elaborate more on partial memory? Isn't it true that if a computer fails, all the information in memory is lost? How can we use partial memory for node recovery?