I am interested in designing, developing, and analyzing novel computer systems. This is mostly driven by my curiosity to find out the implications of expressing newer and useful ideas and pushing the boundaries of what existing computer systems can achieve. Over the course of my Ph.D., I have focused on building systems to manage data, both on mobile systems and on the cloud.
My current research focus is on providing flexible and personalized storage on mobile systems. We are working on building systems that make the data management on mobile systems better. While our primary goal is to provide users the flexibility to manage their data, we are also trying to understand various preferences a user might have when using an app.
We are building these systems on Android and we have published some of these ideas through our work on BlueMountain.
BlueMountain
BlueMountain adds flexibility for data management on mobile systems by proposing an ecosystem which decouples the app-logic and data management logic. BlueMountain proposes a number of practical solutions for seamlessly transitioning from the current eco-system to what we propose.
These extensions were added based on the experiences from our initial work which explored the design space of an app instrumentation tool that automatically integrates cloud storage services for Android apps. Our current system allows a user to download an app from an online app store and customize its data management by linking it with any of the available data management solutions. You can find more information on our project website BlueMountain.
Sharath Chandrashekhara, Taeyeon Ki, Kyungho Jeon, Karthik Dantu, Steven Y. Ko
Proceedings of the 23rd Annual International Conference on Mobile Computing and Networking (MobiCom) 2017
[bibtex] [Paper (PDF)]
Sharath Chandrashekhara, Taeyeon Ki, Kyungho Jeon, Karthik Dantu, Steven Y. Ko
Proceedings of the 15th Annual International Conference on Mobile Systems, Applications, and Services (MobiSys), 2017
[Poster(PDF)] [Abstract (PDF)]
Sharath Chandrashekhara, Kyle Marcus, Rakesh G. M. Subramanya, Hrishikesh S. Karve, Karthik Dantu, Steven Y. Ko
Proceedings of the 7th USENIX Workshop on Hot Topics in Storage and File Systems (HotStorage), 2015
[bibtex] [Paper (PDF)]
Cider
Cider is a highly flexible and scalable virtual block device, supporting temporally variable data redundancy. Cider works as a distributed block data store, using erasure coding for reliability and allows controlling data redundancy at the granularity of a single block.
We built our initial prototype on disk based systems. We have also proposed techniques to integrate our system with distributed flash-based arrays. To test this approach, we have developed a prototype of an SSD simulator which can simulate a system based on Cider's technology. Our goal is to create a centrally managed flash based array, in which some of the low level functionality of FTLs in each SSD is moved to the master node which manages the flash chips spanning across multiple storage nodes. This work has produced a patent (pending) and a conference publication.
Sharath Chandrashekhara, Madhusudhan R. Kumar, Mahesh Venkataramaiah, and Vipin Chaudhary
The 26th International Conference on Computer Communications and Networks (ICCCN 2017).
[bibtex] [Paper (PDF)]
System and method for fault-tolerant block data storage
Inventors: Sharath Chandrashekhara, Madhusudhan Ramesh Kumar, and Vipin Chaudhary
Application status: Patent filed in April 2015, published on October 2015 (Patent pending).
[bibtex] [Patent Claim Document]
Pixelsior
Pixelsior is a photo management framework
which can manage the photos on a device and can
transparently manage content adaptation and image manipulation.
Pixelsior also assists the development of image manipulation apps by
providing a pluggable interface for image filters.
This is a work in progress, led by Kyungho Jeon. This is based
on our initial work which was published in HotStorage, which presents
our study of the requirements of a photo management system and a
preliminary design. A full paper in under submission.
Kyungho Jeon, Sharath Chandrashekhara, Karthik Dantu, Steven Y. Ko
Proceedings of the 8th USENIX Workshop on Hot Topics in Storage and File Systems (HotStorage), 2016
[bibtex] [Paper (PDF)]
PigOut
PigOut is a system that enables federated data processing over multiple Hadoop clusters. You can find more information on our project website PigOut.
Kyungho Jeon, Sharath Chandrashekhara, Feng Shen, Shikhar Mehra, Oliver Kennedy, Steven Y. Ko
Proceedings of the 2014 IEEE International Conference on Big Data (BigData), 2014.
[bibtex] [Paper (PDF)]