There has been research exploration into a topic called "Database processing Units" which are customized ASICs for processing database transactions. They have been touted as a solution for databases.
I guess the era of single-core silicon CPU speed-up is over. Research is currently being carried out to evaluate other materials as Si replacements so that the limitations mentioned in the previous lecture can be addressed.
o_o
Because single-core CPU speedup will not be improved, database applications have to be able to utilize machines that have more and more cores. However, at this point in time most are only optimized for 32 core machines, so they can't take advantage of having more cores.
There has been research exploration into a topic called "Database processing Units" which are customized ASICs for processing database transactions. They have been touted as a solution for databases.
A reference paper can be found here: LINK
I guess the era of single-core silicon CPU speed-up is over. Research is currently being carried out to evaluate other materials as Si replacements so that the limitations mentioned in the previous lecture can be addressed.
Because single-core CPU speedup will not be improved, database applications have to be able to utilize machines that have more and more cores. However, at this point in time most are only optimized for 32 core machines, so they can't take advantage of having more cores.