Scalability with Parallelism{1}


Embarrassingly Scalable Database Systems talks about how scalability will be achieved through parallelism as technology progresses over time. It mentions Moore’s Law and how it translates to single-processor performance gains. Moore’s Law states that the number of processors per chip roughly doubles every two years. The challenge for conventional database servers running business intelligence and transaction processing workloads is parallelism. As the cost of microprocessors decrease, parallelism is going to be used on a larger scale to increase scalability. It mentions that there is a tradeoff when it comes to performance and scalability. It also mentions that in order to transform a database storage manager from a single to a multi-threaded processing, many fundamental changes need to be made. To achieve parallelism on a larger scale many changes need to be made, for example, decoupling transaction data access is used to ensure consistency in centralized operations and other technical changes. It mentions the areas where parallelism can be utilized such as the query processing level.

This article caught my attention with its title. But after reading its content it proved to be very informative. It introduced a very practical problem when it comes to scalability using parallelism. As technology advances and microprocessors become less expensive, it is going to be difficult to split tasks further to maintain efficiency for database processing. Therefore other methods are required to maintain efficiency when processing tasks are divided in larger quantities.

Since parallelism is becoming more and more popular when it comes to large database processing, I found that it was important to cover this article and introduce this issue, and the different methods that are suggested to resolve some of the problems. It also relates to my previous article.

Anastasia Ailamaki. “Embarrassingly Scalable Database Systems, This paper appears in:
Data Engineering (ICDE), 2011 IEEE 27th International: 1. IEEE Xplore Digital Library. 11 April. 2011 – 16 April 2011. Web. 27 May 2012. <http://0-ieeexplore.ieee.org.opac.library.csupomona.edu/stamp/stamp.jsp?tp=&arnumber=5767964>