Current Research

My current research focuses on datapath operating systems for microsecond-scale datacenter systems.

• 
  Paper

  Demikernel

  The Demikernel is a new datapath OS and architecture for microsecond-scale datacenter systems and kernel-bypass devices. Demikernel accommodates heterogenous kernel-bypass devices with a flexible library OS architecture and new high-level datapath API with an asynchronous I/O interface and zero-copy memory semantics for microsecond I/O processing. Demikernel implements this API for RDMA, DPDK and SPDK devices with new nanosecond-scale I/O stacks in Rust. Once ported to Demikernel, microsecond datacenter systems can run across different devices with no code changes.
  
• 
  Paper

  Treehouse

  Treehouse is new project that proposes a software-centric approach to green datacenters. While existing research aims to reduce energy usage of datacenter hardware, we argue that being energy-aware in the design of datacenter software is equally important. To that end, Treehouse aims to make energy and carbon visible to application developers on a fine-grained basis, modify system APIs to make informed trade offs between performance and carbon emissions, and raise the level of application programming for flexible use of more energy efficient means of compute and storage. Find out more about the Treehouse project at our project webpage!
  

Past Research

My PhD research broadly covered distributed systems for large-scale, wide-area applications. Some recent projects are listed below.

• 
  Paper

  Marvin

  Marvin is a new memory manager for Android that co-designs the application runtime (i.e., JVM) and operating system. Marvin’s key insight is that all mobile applications run within the same runtime, so we can leverage runtime insight into the application for better resource management. Marvin implements almost all memory management in a modified Android Runtime to achieve better memory utilization and performance.
  
• 
  Paper Talk Code More

  TAPIR

  The Transactional Application Protocol for Inconsistent Replication (TAPIR) is a new protocol for linearizable distributed transactions built using replication with no consistency guarantees. By enforcing consistency only at the transaction layer, TAPIR eliminates coordination at the replication layer, enabling TAPIR to provide the same transaction model and consistency guarantees as existing systems like Spanner with better latency and throughput.
• 
  Paper Talk Code More

  Diamond

  Diamond is a new data management system for wide-area, reactive applications. Reactive applications give users the illusion of continuous synchronization across mobile devices and the cloud server. Diamond simplifies this task by providing applications with persistent cloud storage, reliable synchronization between storage and mobile devices, and automated execution of application code in response to shared data updates.