No sql Summary:
shared nothing architecture
compression technology is widely used
store records as row/column
parallell processing (MPP): MapReduce
MapReduce: lack of schema, prevent perfomance improvment from B-tree and hash partition;
static data; small dataset, lagre dataset, considering read/write/insert efficiency
topic:
Netazza: move query close data vs move data close query (proces
Vertica: compression and replicate WOS and ROS: write optimized store and read optimized store
ESAsolution: dynamic index
Rdbms Summary: recovery system is not based on log, but copies.
