Bio


I am a first year Phd Student at the University of British Columbia in the NSSL lab. I work under the supervision of Andrew Warfield.

My research investigates the design and implementation of network available storage systems, particularly with respect to enterprise environments.

Projects


Filesystem study: Most evaluations of deduplication rates are small in scale or from highly self-selecting users. This, combined with an interest in deduplication as primary storage, and the lack of recent filesystem metadata studies led us to perform a large scale study of windows filesystem content and structure at Microsoft.

A Study of Practical Deduplication. Dutch T. Meyer, William J. Bolosky. FAST 2011 [pdf]
Awarded Best Paper

Capo: We performed a deep and detailed analysis of filesystem activity in a production Virtual Desktop Infrastructure (VDI) system at UBC. Based on these findings, we propose and evaluate Capo, which uses caching, multicast, and novel consistency guarantees based on the filesystem namespace to lower peak storage system loads.

Capo: Recapitulating Storage for Virtual Desktops. Mohammad Shamma, Dutch T. Meyer, Jake Wires, Maria Ivanova, Norman C. Hutchinson, and Andrew Warfield. FAST 2011 [pdf]

Parallax: Virtual block storage designed for virtual machines. Provides fine-grained copy-on-write storage and unlimited, very lightweight snapshotting. Project web page

Parallax: Virtual Disks for Virtual Machines. Dutch T. Meyer, Gitika Aggarwal, Brendan Cully, Geoffrey Lefebvre, Mike Feeley, Norm Hutchinson and Andrew Warfield. EuroSys 2008. [pdf]

Dovetail: Storage systems upgrades are often avoided because they are slow, complicated, and dangerous. Dovetail is a storage system upgrade framework, deployed in a real service at UBC, that supports lossless software rollback and protection to make software upgrades easy.

Fast and Cautious Evolution of Cloud Storage. Dutch T. Meyer, Mohammad Shamma, Jake Wires, Quan Zhang, Norman C. Hutchinson and Andrew Warfield. HotStorage 2010. [pdf]

Remus: Virtualization-based invisible, comprehensive disaster protection for commodity operating systems and applications. Project web page

Remus: High Availability via Asynchronous Virtual Machine Replication. Brendan Cully, Geoffrey Lefebvre, Dutch T. Meyer, Mike Feeley, Norm Hutchinson and Andrew Warfield. NSDI 2008.[pdf]
Awarded Best Paper

Plastic: False sharing can cause dramatic performance collapse due to workload-, compiler-, or hardware-based characteristics that are sometimes difficult to predict. We present the first system for transparently detecting and correcting false sharing in an Operating System.

Whose Cache Line Is It Anyway? Operating System Support for Live Detection and Repair of False Sharing . Mihir Nanavati, Mark Spear, Nathan Taylor, Shriram Rajagopalan, Dutch T. Meyer, William Aiello, and Andrew Warfield. Eurosys 2013.[pdf]

Block Mason: Although the block interface is simple, block devices themselves must generally be implemented in the operating system kernel, an environment which is neither simple nor portable. Block Mason allows users to build small, reusable block processing elements in user space, and to connect them together into powerful composite modules using a simple declarative language.

Block Mason. Dutch T. Meyer, Brendan Cully, Jake Wires, Norman C. Hutchinson and Andrew Warfield. WIOV 2008. [pdf]

Windows Tracing Software: I've had some requests for the driver I wrote to trace windows clients in the Capo paper, because people need an example of working in the windows kernel, have some unusual behavior to diagnose, or are doing a filesystem study. I've made this code available under a BSD license, I hope I will hear from people that it's useful

Code is available as a: .zip