Saptaparni Kumar: Fault-tolerant atomic storage in truly-dynamic message-passing systems
Distributed computing is ubiquitous today ranging from multiprocessors to smart devices to data centers connected by small LANs and massive WANs like the Internet. My research focuses on fault-tolerant distributed storage services in the middleware layer of a dynamic distributed system.
A shared register is a fundamental service used by middleware in a distributed system that stores a value and has two operations: read, which returns the value stored in the register, and write, which updates the value stored. Registers serve as building blocks for more powerful data structures. I implement atomic shared registers on top of faulty (Crash faults and Byzantine faults) dynamic message-passing systems, where nodes enter and leave continuously thus allowing the system to grow and shrink arbitrarily.