Presentation One: Towards Predictable Datacenter Networks.ppt
Presentation Two: Accountable Virtual Machines.ppt
Project Introduction and Goals
With the rapid growth of information and data, many areas such as science, engineering and business computing need to deal with large-scale data. As the most important technology in Cloud Computing, virtualization allows the applications to run encapsulated inside a VM. However, it limits opportunities for data reuse as each VM keeps it own independent buffer cache, and result in wasteful duplication if multiple VMs that operate on overlapping files are collocated on the same host. Octopus-File System (OFS) enables cross-VM buffer cache sharing. So the goal of this project is to give the comparison between NFS and OFS through the execution of scheduler built on top of OFS to illustrate the good performance of OFS and find the best algorithm to reduce amount of network transfer, decrease the load on the file server, and increase the system throughput.
Sep 26 - Oct 3
Built experimental environment
Oct 4 - Oct 17
Characterize applications runtime
|Oct 18 - Oct 24||Let scheduler running on the machines applied from EC2|
|Oct 24||The First Progress Report|
|Oct 25 - Oct 31||Configure a pool of NFS servers|
|Nov 7 - Nov 21||Compare NSF and OFS|
|Nov 21||The Second Progress Report|
|Nov 22 - Nov 28||Finishing the experimental results|
|Nov 29 - Dec 5||Finishing the summary and analysis of the experiment|
|Dec 8 - Dec 19||
Finishing the final report
 Svitlana Tumanova, Octopus: Cache-Based Data-Local Scheduler (manuscript).
 Isard, M., Prabhakaran, V., Currey, J., Wieder, U., Talwar, K., and Goldberg, A. Quincy: fair scheduling for distributed computing clusters. In Proceedings of the ACM SIGOPS 22nd symposium on Operating systems principles (New York, NY, USA, 2009), SOSP ¡¯09, ACM, pp. 261¨C276.
 BLAST: Basic Local Alignment Search Tool. http://blast.ncbi.nlm.nih.gov/Blast.cgi, 2011.
 Zaharia, M., Borthakur, D., Sen Sarma, J., Elmeleegy, K., Shenker, S., and Stoica, I. Delay scheduling: a simple technique for achieving locality and fairness in cluster scheduling. In Proceedings of the 5th European conference on Computer systems (New York, NY, USA, 2010), EuroSys ¡¯10, ACM, pp. 265¨C278.