A PHD POSITION IS AVAILABLE NOW at the Department of Electrical & Computer Systems Engineering, Monash University, Australia!
Topic : Coding theory techniques in failure recovery for data storage systems
Duration : 2019-2022
Scholarship : 27,353 AUD/year (if the student also manages to get the Monash scholarship, there’ll be a top up to the salary to cover other expenses like health insurance, relocation, etc.)
Supervisor : Prof. Emanuele Viterbo – IEEE Fellow ( https://ecse.monash.edu/staff/eviterbo/ )
Co-supervisor : Dr. Son Hoang Dau – DECRA Fellow ( https://sites.google.com/site/dausonhoang84/ )
Deadline : from now until the position is filled
Contact : interested students may send a CV and color scans of his/her Bachelor/Master degrees and transcripts to Dr. Hoang Dau at [email protected]
Requirements
Background in Mathematics, Computer Science, Electrical Engineering, or closely related areas
- Solid knowledge of Linear Algebra
- Knowledge of Discrete Mathematics / Abstract Algebra / Cryptography is a plus
- Comfortable with one of the major programming languages: C/C++/Java/Python
- English: IELTS 6.5 overal, or equivalent (in some exceptional cases, an excellent student with a lower score can still be accepted, but extra training on English is required upon arrival in Australia)
- Other requirements from Monash University: https://goo.gl/UyDHQv
Project summary
Distributed storage systems form an essential component in every contemporary data centre, guaranteeing data availability, scalability, and cost effectiveness. A distributed storage system is formed by networking a large number of storage servers (a.k.a. storage nodes), which are located at different racks within a single data center or even across geographically dispersed data centers around the globe. In order to avoid data loss and to increase data availability, each stored file is split into several data blocks, which are then replicated, or more often, transformed into coded blocks using some erasure codes, e.g. Reed-Solomon codes. These blocks are then distributed to different storage nodes across the system. A user can recover the file by retrieving coded blocks stored at any suitable set of storage nodes.
Replication and erasure codes guard the stored data against software and hardware failures by adding an appropriate level of redundancy. Erasure codes are a favourable choice due to their lower storage overhead, their much higher read/write throughputs thanks to parallelism, and their higher mean-time-to-failures, compared to replication. Indeed, erasure codes are currently employed by a number of prominent companies such as Google, Facebook, Baidu, Yahoo, Backblaze, Amazon, and Microsoft, to protect their storage systems.
The project addresses fundamental theoretical questions about the structure of erasure codes, in particular, Reed-Solomon codes, with respect to their repair capability and limitation, as well as tackles the practical problem of improving the recovery performance of distributed storage systems. Key performance criteria include but not limited to the bandwidth, I/O, and computation overhead required during the recovery process of erasure codes.
More information about this project can be found online at https://goo.gl/HSKtEg .
Monash was ranked number one in Australia and 45th globally in engineering by the Times Higher Education World University Rankings 2016-2017. The Faculty of Engineering at Monash is regarded as performing research well above the world standard (ERA: 5) in both 2012 ERA and 2015 ERA.
The Department of Electrical and Computer Systems Engineering (ECSE) currently hosts three distinguished IEEE Fellows, Professors Jean Armstrong, Arthur James Lowery and Emanuele Viterbo. Professor Viterbo is a leading expert in the fields of information theory and coding theory, and serves as an editor for high-impact journals in the fields, including the IEEE Transactions on Information Theory and Foundations and Trends in Communication and Information Theory.