Swarthmore College Department of Computer Science

Talk by Charlie Garrod, Swarthmore College CS

Scalable, Efficient Distributed Data Services
Thursday, February 9, 2012
SCI 240, 4:00 pm (cookies at 3:45)

Abstract

Relational databases are a key component in many software systems, yet they are not ideal for many of the tasks for which they are used. In this talk I will examine how typical modern web applications use data, discuss why databases are poorly suited for those tasks, and propose one alternative: distributed data management services.

Distributed data management services are reliable, scalable distributed components that can be used to build more-complex distributed systems. Their interface is modeled after familiar data structures we all use, such as sets. I'll describe one possible architecture to implement distributed data management services -- using scalable key-value systems as the underlying storage mechanism -- and discuss advantages and disadvantages of that implementation.