A major focus of my work has been flash storage. This is the first new storage technology in most of our lifetimes which offers serious competition to magnetic disk, and while researchers have spent decades learning studying disk performance and how to optimize it, little is yet known about flash behavior and how to best use it. Much of my work in this area focuses on performance of Flash Translation Layer (FTL) algorithms - i.e. the question of how to predict write amplification, which in turn determines performance and lifetime.
Shingled Magnetic Recording (SMR) disk drives
SMR drives are a new class of disk, which gain higher capacity by sacrificing the ability to overwrite disk sectors in place. This in turn forces the use of log-structured or out-of-place write algorithms much like those used for Flash.
For the last two years I have been partnering with Orran Krieger at BU in the development of the Massachusetts Open Cloud, a partnership between five universities and a large number of industry partners which is establishing a *multi-provider* public cloud based in the MGHPCC data center in Holyoke.
If you are a talented student who is interested in any of these areas, please send me an email to see if there are any opportunities to work together.