Publication

• DeltaFS IMD
  Scaling Embedded In-Situ Indexing with DeltaFS (SC18) [slides]
• DeltaFS IMD
  Software-Defined Storage for Fast Trajectory Queries using a DeltaFS Indexed Massive Directory (PDSW17) [slides]
• SlimDB
  A Space-Efficient Key-Value Storage Engine For Semi-Sorted Data (PVLDB17)
• DeltaFS
  Exascale File Systems Scale Better without Dedicated Servers (PDSW15) [slides]
• ShardFS
  Replication vs. Caching Strategies for Distributed Metadata Management in Cloud Storage Systems (SOCC15) [slides]
• BatchFS
  Scaling the File System Control Plane with Client-Funded Metadata Servers (PDSW14) [slides]
• IndexFS
  Scaling File System Metadata Performance with Stateless Caching and Bulk Insertion (SC14 Best Paper Award) [slides]
• COSBench
  Cloud Object Storage Benchmark (ICPE13) [github]
• COSBench
  A Benchmark Tool for Cloud Object Storage Services (CLOUD12)

Indexed Massive Directories

• Quickly absorb 1 trillion files while simultaneously allowing each of them to be read efficiently. How to transform a filesystem directory to a record-breaking hero for national lab scientists (please see our DeltaFS IMD paper for details)

Scalable Filesystem Metadata Designs

• No more false sharing: filesystems scale better with no dedicated metadata servers (please see our DeltaFS paper for details)
• Orders of magnitude faster than the state-of-the-art: redesigning filesystems with 1) client-funded metadata servers, 2) aggressive client-side metadata buffering, and 3) self-provable metadata mutation logs (please see our BatchFS paper for details)
• Distributed filesystem metadata made better with 1) column-based metadata representation, 2) efficient client-side metadata caching, and 3) fast metadata bulk insertion (please see our IndexFS paper for details)
• Scalable namespace partitioning with efficient pathname lookups: will server-side replication outperform client-side caching? (please see our ShardFS paper for details)

Cloud Object Storage

• Checkout this awesome cloud object storage benchmark tool COSBench (https://github.com/intel-cloud/cosbench) that enables better performance characterization, comparison, and optimization (please see our COSBench paper for details)