Survey on Network Storage Systems with Flash Memory

PhD Qualifying Examination


Title: "Survey on Network Storage Systems with Flash Memory"

by

Mr. Haoyu TAN


Abstract:

The last decade has seen an explosive development of data-intensive
applications.

Both scientific and industrial community are looking for a way to manage 
peta-scale or even exa-scale data. Today, many different categories of 
systems such as distributed file systems (GFS, Luster) data-intensive 
computing platform (MapReduce, Amazon EC2), and peta-scale databases 
(SciDB, Bigtable) are all aimed at large-scale data storing and 
processing. However, large-scale network storage system is the basic 
infrastructure for all these higher-level systems. The storage 
architecture can affect the performance of those systems in many ways. For 
example, dedicatedly designed network topology could well serve the 
one-to-many and many-to-many network traffic patterns generated by 
MapReduce, but traditional network topologies may not.  There are various 
requirements for a network storage system to support diverse applications.

Among them include scalability, reliability (fault tolerance), power 
efficiency, manageability and so on. The evolution of flash memory leads 
to a whole new era of building high-performance and power efficient 
storage systems.  Moreover, today many researchers draw attention to 
creative network topology in data centers, such as fat-tree, DCell and 
BCube, which is closely related to the scalability, reliability and 
manageability of maintaining thousands of storage nodes. In this survey, 
we are going to address the development of these two technologies, and 
discuss how they will affect the design of network storage systems with 
flash memory.


Date:			Tuesday, 23 February 2010

Time:			10:00am - 12:00noon

Venue:			Room 3301
			lifts 17/18

Committee Members:	Prof. Lionel Ni (Supervisor)
			Dr. Qiong Luo
			Dr. Lei Chen
			Dr. Lin Gu


**** ALL are Welcome ****