Tashi: Cloud Computing on Big Data

Tashi is a research project at Intel Labs Pittsburgh designed to investigate the implications of cloud computing on big data. As part of this effort, researchers at Intel Labs Pittsburgh are also contributing to the development of an open source cluster management system from the Apache Software Foundation's (ASF) Incubator, also called Tashi (the initial proposal). Key initial contributors include Intel Labs Pittsburgh and the Parallel Data Laboratory at Carnegie Mellon University.

The Tashi cluster management system leverages virtual machine technology to enable deployment of many virtual clusters with differing software requirements on a single physical cluster, and a major research goal of the project is the development of mechanisms that will enable efficient access to cluster resources, such as power and storage, from these independent virtual clusters.

Tashi is deployed on the Open Cirrus cluster at Intel Labs Pittsburgh. This 200-node+ cluster comprises more than 1500 cores and 750 disks (providing more than 0.5 PB of raw storage). Tashi is a key software component that enables the Big Data cluster to participate in the OpenCirrus cluster testbed.

Intel Labs Pittsburgh Team

---

• Michael Kozuch (PI)
• Richard Gass
• David O'Hallaron

Collaborators

---

• Greg Ganger (CMU)
• Garth Gibson (CMU)
• Mor Harchol-Balter (CMU)
• Julio López (CMU)
• Michael Stroucken (CMU)
• Jim Cipar (CMU)
• Anshul Gandhi (CMU)
• Tomer Shiran (CMU)
• Wittawat Tantisiriroj (CMU)
• Michael P. Ryan

Publications

---

• "Cloud Computing on Rich Data," Michael Kozuch, Jason Campbell, Madeleine Glick, Padmanabhan Pillai, Intel Technology Journal, Volume 14, Issue 1, 2010, Pages 114-127. [link]
• "Optimality analysis of energy-performance trade-off for server farm management," Anshul Gandhi, Varun Gupta, Mor Harchol-Balter, Michael A. Kozuch, Performance Evaluation, Volume 67, Issue 11, Performance 2010, November 2010, Pages 1155-1171.
• "Robust and Flexible Power-Proportional Storage," Hrishikesh Amur, James Cipar, Varun Gupta, Michael Kozuch, Gregory Ganger, Karsten Schwan, Symposium on Cloud Computing (SOCC), June 2010.
• "Cluster Fault-Tolerance: An Experimental Evaluation of Checkpointing and MapReduce through Simulation," Thomas C. Bressoud and Michael A. Kozuch, 2009 IEEE International Conference on Cluster Computing (Cluster'09), September 2009.
• "Tashi: Location-aware Cluster Management", Michael A. Kozuch, Michael P. Ryan, Richard Gass, Steven W. Schlosser, David O'Hallaron, James Cipar, Elie Krevat, Julio López, Michael Stroucken, Gregory R. Ganger, First Workshop on Automated Control for Datacenters and Clouds (ACDC'09), June 2009.
• "Migration without Virtualization", Michael A. Kozuch, Michael Kaminsky, Michael P. Ryan, Workshop on Hot Topics in Operating Systems (HotOS '09), May 2009.
• "Tashi: Cloud Computing on Big Data", PDL Packet: Newsletter on PDL Activities and Events, (Adapted from ACDC'09), [pdf].