Research

  • ML for Systems and Systems for ML
    • LMSys (e.g., LangChain and FlashAttention)
    • RAGS and VectorDB
    • Data lake analytics using multi-modality models (e.g., VLM)
  • Cloud, Distributed, Database, Operating Systems, and HPC (Supercomputing)
    • Distributed/transactional protocols
    • Serverless, and disaggregated memory on CXL
    • GPGPU, lock-free data structures, and many-core synchronization
    • Learned indexing and concurrency control

The followings are some of my ongoing projects.


GeoDB
Geolocated databases shard and replicate data across continents for scalability and availability.  In this project, we are building a new one that has both scalability and availability without sacrificing consistency, with optional security concerns.
  • Status: New
  • Area: Distributed systems × Database
  • Keywords: Sharding, Replication, Consensus, Concurrency Control
  • Selected Publications:
    • Ziliang Lai, H Fan, W Zhou, Z Ma, X Peng, Feifei Li, Eric Lo: “Knock Out 2PC with Practicality Intact: a High-performance and General Distributed Transaction Protocol”.  ICDE, 2023.
    • Ziliang Lai, Chris Liu, Eric Lo: “When Blockchain meets Deterministic Database”.  SIGMOD, 2023.
    • Mohammad Javad Amiri, Ziliang Lai, Liana Patel, Boon Thau Loo, Eric Lo, Wenchao Zhou: “Saguaro: Efficient Processing of Transactions in Wide Area Networks using a Hierarchical Permissioned Blockchain”. ICDE, 2023.

Fast.AI / Edge.AI
We are in the midst of deep AI.  While deep models (e.g., large language models, and deep vision models) achieve unprecedented performance, using them for high-throughput applications (e.g., analyzing a large volume of videos) or on the edge (e.g., when the video cannot leave the source for privacy reasons) can be very expensive or inefficient (imagine running LLM locally on your cell phone).  In this project, we investigate various techniques to enable fast model inference on various platforms and hardware.
  • Status: Ongoing
  • Areas: MLSys
  • Keywords: Deep Model Inference, MLOps
  • Selected Publications:
    • PengFei Zhang, C. Han, Eric Lo: “More is Less – Byte-quantized models are faster than bit-quantized models on the edge”.  In Proceedings of BigData, 2022.
    • PengFei Zhang, Eric Lo, Baotong Lu: “High-Performance Depthwise and Pointwise Convolutions on Mobile Devices”.  In Proceedings of AAAI, 2020.
    • Ziliang Lai, Chenxia Han, Chris Liu, PengFei Zhang, Eric Lo, Ben Kao: “Top-K Deep Video Analytics: a Probabilistic Approach“.  SIGMOD, 2021. Also check out project web.

Architectural Conscious Data Processing

Continued to thrive for ever-faster processing is leading computer scientists to directly leverage modern hardware innovations in the design of software systems. This trend is further amplified by the collapse of improvements in linear chip clock frequency scaling due to physical limits.  Therefore, software system designs that disregard hardware innovations are doomed to failure.

Our goal is to study different approaches to leveraging modern/emerging hardware to accelerate data processing in big data management.  Hardware under consideration includes many-core processors (e.g., GPGPU), Disaggregated Memory, CXL, etc.

  • Status: Ongoing
  • Areas: Database × Operating Systems × Supercomputing
  • Keywords: Disaggregated Memory, SIMD, RDMA, many-core, GPGPU, software-hardware co-design
  • Selected Publications:
    • Chaichon. Wongkham, B. Lu, C. Liu, Z. Zhong, Eric Lo, T. Wang: “Are Updatable Learned Indexes Ready?“. In Proceedings of VLDB, 2022.
    • B. Lu, J. Ding, Eric Lo, U.Farooq Minhas, T. Wang: “APEX: A High-Performance Learned Index on Persistent Memory“. In Proceedings of VLDB, 2022.
    • B. Lu, X. Hao, T. Wang, Eric Lo: “Dash: Scalable Hashing on Persistent Memory“.  In Proceedings of VLDB, 2020. (ACM SIGMOD Research Highlight Award 2021)
    • W. Xu, Eric Lo, P. Zhang: “DIFusion: Fast Skip-Scan with Zero Space Overhead“.  In Proceedings of IEEE ICDE Conference, 2018.
    • W. Xu, Z. Feng, Eric Lo: “Fast Multi-column Sorting in Main-Memory Column-Stores“.  In Proceedings of ACM SIGMOD Conference, 2016.
    • Z. Feng, Eric Lo, B. Kao, W. Xu: “ByteSlice: Pushing the Envelop of Main Memory Query Processing with a New Storage Layout“.  In Proceedings of ACM SIGMOD Conference, 2016.
    • Z. Feng, Eric Lo: “Accelerating aggregation using intra-cycle parallelism“.  In Proceedings of IEEE ICDE Conference, 2015.