Multiprocessing Scalability

by David L
In this weeks article by Anastasia Ailamaki, named “Embarrassingly Scalable Database Systems”, she talks about how the future trend is favoring multiprocessing. She says “A careful analysis of database performance scaling trends on future multiprocessors demonstrates that the current parallelism are of bounded utility as the number of processors per chip increases exponentially” (Ailamaki, 2011). She also gives an example comparing two Greek mythological figures, Atlas and the Hydra. Atlas would be the powerful single-threaded processor, and the Hydra would be a multi-threaded processor that could scale to infinity. But a key point she notes is that in order to process some processes are designed to be executed by a central operation through parallelism we must make sure that the process is consistent when decoupling the data(Ailamaki, 2011)

This relates to our database class because this article talks about how the database processes information, and database performance. In the future, when parallel processing can outperform most of the “central operations”, we could see a future leading to an extremely broad range of multicore processors. Even now when we are shopping online or in electronic stores, we are able to purchase the mainstream single, dual, quad, or hexacore processors, but not too long ago, Intel did state that they had an 80 core processor. In the future, think about how much more things we can accomplish/get done with a high number of multicore processors. We could make games even more realistic, fold protein molecules faster, and process other data intensive tasks easier.


Anastasia Ailamaki. “Embarrassingly Scalable Database Systems” Data Engineering (ICDE), IEEE 27th International Conference (2012). IEEE Xplore Digital Library. 11 April. 2011 – 16 April 2011. Web. 28 May 2012. <>