View Rohan Gandhi's profile on LinkedIn

Rohan Gandhi

I am currently a post-doc fellow at Carnegie Mellon University with Prof. Vyas Sekar. Prior, I completed my Ph.D. from Purdue University advised by Prof. Y. Charlie Hu. I am broadly interested in computer systems and networks.

Projects

    Load balancer trilogy (Duet, Rubik, Yoda)

    Using commodity data-center switches (instead of servers) to load balance network traffic that reduces the load balancer cost and improves latency by 12-25x while providing untethered connectivity despite network failures using a novel way of decoupling the TCP stack, and storing it persistently.

    CoFlow scheduling trilogy (Graviton, Saath, Philae)

    Integrating the spatial dimension of CoFlows (flows on many ports) to improve the CoFlow scheduling by up to 4x.

    Network Update (Dionysus, Catalyst): Dynamic scheduling of the network updates

    A system for fast and consistent network updates in SDN that improves the update speed by 53-88%.

    Pikachu: large-scale data processing

    Hadoop statically partitions the data (to reducer) based on the number of slots. This design choice leads to imbalance in the amount of data (especially on heterogeneous clusters), that elongates job completion time. In Pikachu, we articulate a run-time mechanism that dynamically adjusts the data to different reducer nodes in proportion to their processing speed. Various design choices in Pikachu improve the job-completion time by 25-42% compared to Hadoop.

    Mercury: in-memory key-value store

    Mercury = Most scalable version of Memcached! We re-designed Memcached internals to remove/reduce all bottlenecks in the memory sub-system of Memcached. Mercury achieves near linear speedup with number of cores on any work-load. Compared to Memcached 1.4.13, Mercury processes 4-12x more queries/sec.

    Wireless (Percy) and cognitive radio (DCH)

    In Percy, we improved the video streaming by using layered videos (MDC/MRC), coupled with the network coding. Percy improves video quality by up to 5 PSNR. In DCH, we deviced two algorithms that achieve rendezvous faster by 80%.

Publications

Experience

    Research Intern

  • Microsoft Research, Redmond with Ming Zhang and Jitu Padhye, March'13 - March'14, Dec'14 - Dec'15.
  • IBM Research, Almaden with Anna Povzner and Wendy Belluomini, May'12 - Aug'12.
  • Research Assistant

  • Distributed System and Networking Lab, Purdue University, Jan'11 - May'16
  • Teaching Assistant

  • ECE469: Operating Systems, Purdue University, Spring'14
  • ECE270: Intro. to Digital Design, Purdue University, Fall'10 and Spring'11
  • Application Engineer

  • Cypress Semiconductor, Bangalore and San Jose, May'09 - May'10

Adopted from Polo Chau and Andreas Viklund