[Zubair Nabi]
I am a graduate student in the Computer Science Department at LUMS. I am also both a Researcher and a Teaching Assistant. For my research I work with Dr. Umar Saif and Dr. Atif Alvi. I am the TA of a graduate level course on Advanced Operating Systems this Fall.
My research interests include (but are not limited to) data intensive computing, distributed systems, operating systems, and datacenter networks. I am also interested in ICTD. My master's thesis is based on an improved version of Google MapReduce. It addresses several fundamental weaknesses in the MapReduce architecture, especially when input data has structure or intermediate results have skew.
Current Research Projects
Knocking Down the MapReduce Brick-wall
We have developed a modified architecture of MapReduce, dubbed MR+, which advocates a departure from the fixed two-staged architecture of MapReduce to a flexible, multi-staged implementation. MR+ has several inherent advantages over traditional MapReduce: (1) it is resilient to skew in intermediate results, (2) it avoids the wholesale copying of intermediate data at the end of the map phase which may otherwise paralyze the entire cluster while reduce workers are being loaded with data en masse, (3) it naturally avoids the reduce straggler problem due to a heterogeneous cluster, (4) it may be used to prioritize the processing of large datasets by detecting clusters of useful information in the input data, and (5) it enables early estimation of results for very large datasets. Our improvements over the original architecture still maintain the clean, convenient programming model of MapReduce. Our evaluation shows that MR+ outperforms Hadoop MapReduce, Hadoop Online Prototype and LATE by up to a factor of 7 for our target applications and datasets.
Modern vehicles are equipped with hundreds of embedded networked components with computational, sensory and actuation powers. Reliable functioning and interaction of these components is vital for the safety of the vehicle and its passengers. Our research goals include, formally verifying the compatibility of each component with the rest of the system, providing means to define overall behaviour by using first order logic rules in an ontological space, enabling feature interaction checks in AUTOSAR, using TinyOS to define a common software stack for on-board sensors and implementing a fire-wall to keep external access to the on-board ECU bus in check.
Publications
- Humayum, A., Azam, M., Nabi, Z., and Saif, U., "MR+: Knocking Down the MapReduce Brick-wall", Under Submission: USENIX ATC '11.
- Nabi, Z., and Saif, U., "MR+: Knocking Down the MapReduce Brick-wall", Master's Thesis: Lahore University of Management Sciences, December '10.
- Nabi, Z., Alvi, A., and Mehmood, R., "Towards Standardization of In-Car Sensors", Accepted for Publication: 3rd International Workshop on Communication Technologies for Vehicles (Nets4Cars 2011).
- Alvi, A., Nabi, Z., Greaves, D., and Mehmood, R., "Enabling and Verifying Mobile Learning Pervasive Environments with Knowledge Bases", Under Submission: Journal for Engineering Applications of Artificial Intelligence (EAAI), Elsevier.
- Alvi, A., Nabi, Z., Greaves, D., and Mehmood, R., "Intra-Vehicular Verification and Control: A Two- Pronged Approach", Accepted for Publication: International Journal of Autonomous and Adaptive Communications Systems (IJAACS).
- Alvi, A., Nabi, Z., Greaves, D., and Mehmood, R., "Controlling Real World Pervasive Environments with Knowledge Bases", Knowledge-Based and Intelligent Information and Engineering Systems, Lecture Notes in Computer Science, 2010, Volume 6279/2010, 576-585.
Contact
Neighbourhood for Emerging World Technologies (NEWT)Computer Science Department
Lahore University of Management Sciences
Lahore, Pakistan
Email: zubair(dot)nabi(at)lums(dot)edu(dot)pk
Phone: +92 (42) 111 115 867, Ext: 4156
