About

Michael J. Freedman is the Robert E. Kahn Professor of Computer Science at Princeton University. He is also a co-founder and CTO of Tiger Data SNS Group and an associate of CITP at Princeton. His research interests include distributed systems, networking, and security. The information provided highlights his academic position, his entrepreneurial role, and his areas of research focus, emphasizing his contributions to these fields within the context of his work at Princeton University.

Research topics

• Computer Science
• Artificial Intelligence
• Distributed computing
• Algorithm
• Computer hardware
• Computer network
• Parallel computing
• Business
• Psychology
• Operating system

Selected publications

• Fusion: An Analytics Object Store Optimized for Query Pushdown

  2025-02-06 · 3 citations

  articleOpen accessSenior author

  The prevalence of disaggregated storage in public clouds has led to increased latency in modern OLAP cloud databases, particularly when handling ad-hoc and highly-selective queries on large objects. To address this, cloud databases have adopted computation pushdown, executing query predicates closer to the storage layer. However, existing pushdown solutions are inefficient in erasure-coded storage. Cloud storage employs erasure coding that partitions analytics file objects into fixed-sized blocks and distributes them across storage nodes. Consequently, when a specific part of the object is queried, the storage system must reassemble the object across nodes, incurring significant network latency.

  PublisherOA PDFDOI

• Using Cryptography to Ensure Integrity in Distributed Ledger Technology

  American Journal Of Cryptography And Network Security · 2025-12-31

  article1st authorCorresponding

  Distributed Ledger Technology (DLT) has revolutionized how digital transactions are recorded and verified, enabling decentralized trust without intermediaries. Cryptography plays a foundational role in ensuring the integrity of data within these ledgers by providing mechanisms for secure data hashing, digital signatures, and consensus protocols. This paper explores the cryptographic techniques essential to maintaining integrity in DLT, including hash functions, public-key cryptography, and Merkle trees. We analyze how these tools mitigate tampering and fraud while supporting transparency and immutability in distributed environments. A comparative overview of cryptographic primitives in popular blockchain platforms is also provided. The study concludes by highlighting future challenges and potential advancements in cryptographic methods for enhancing DLT integrity.

  PublisherDOI

• Efficient Compactions between Storage Tiers with PrismDB

  2023-03-20 · 8 citations

  articleOpen accessSenior author

  In recent years, emerging storage hardware technologies have focused on divergent goals: better performance or lower cost-per-bit. Correspondingly, data systems that employ these technologies are typically optimized either to be fast (but expensive) or cheap (but slow). We take a different approach: by architecting a storage engine to natively utilize two tiers of fast and low-cost storage technologies, we can achieve a Pareto efficient balance between performance and cost-per-bit.

  PublisherOA PDFDOI

• Uniqueness in Haken’s Theorem

  The Michigan Mathematical Journal · 2022-12-19 · 3 citations

  article1st authorCorresponding

  Following Haken [Ha] and Casson and Gordon [CG], it was shown in [Sc] that given a reducing sphere or ∂-reducing disk E in a Heegaard split manifold M, the Heegaard surface T can be isotoped so that it intersects E in a single circle. Here we show that when this is achieved by two different positionings of T, one can be moved to the other by a sequence of ∙ isotopies of T rel E, ∙ pushing a stabilizing pair of T through E, and ∙ eyegelass twists of T. This last move is inspired by one of Powell’s proposed generators for the Goeritz group [Po].

  PublisherDOI

• Non-separating immersions of spheres and Bing houses

  Mathematical Research Letters · 2021-01-01

  preprintOpen access1st authorCorresponding

  We construct Bing houses in all dimensions $n \geq 3$, obtaining non-separating PL immersions of $\mathbb{S}^n \rightarrow\mathbb{R}^{n+1}$.

  PublisherOA PDFDOI

• Approximation rigidity and $h$-principle for Bing spines

  Mathematical Research Letters · 2021-01-01

  preprintOpen access1st authorCorresponding

  We show that all PL manifolds of dimension $\geq 3$ have spines similar to Bing's house with two rooms. Beyond this we explore approximation rigidity and an $h$-principle.

  PublisherOA PDFDOI

• A Glossary of Terms Used in Educational Assessment

  2021-09-03 · 1 citations

  book-chapter1st authorCorresponding
  PublisherDOI

• Efficient Migrations Between Storage Tiers with PrismDB.

  arXiv (Cornell University) · 2021-09-23

  preprintOpen accessSenior author

  In recent years, emerging hardware storage technologies have focused on divergent goals: better performance or lower cost-per-bit of storage. Correspondingly, data systems that employ these new technologies are typically optimized either to be fast (but expensive) or cheap (but slow). We take a different approach: by designing a storage engine to natively utilize two tiers of fast and low-cost storage technologies, we can achieve a Pareto-efficient balance between performance and cost-per-bit. This paper presents the design and implementation of PrismDB, a novel key-value store that exploits two extreme ends of the spectrum of modern NVMe storage technologies (3D XPoint and QLC NAND) simultaneously. Unlike prior work that has retrofitted data structures designed for a single tier to multi-tier storage, PrismDB's data structures and migration mechanism are tailored for different storage layers. The key design contributions of PrismDB is a novel hybrid data structure for tiered storage, and an adaptive and lightweight migration mechanism that is able to efficiently demote cold objects from the fast to the slow storage tier, and promote hot objects to the fast tier. Compared to the standard use of RocksDB on flash in datacenters today, PrismDB's average throughput on tiered storage is 3.3$\times$ faster and its read tail latency is 2$\times$ better, using equivalently-priced hardware.

  Publisher

• VirtualFlow: Decoupling Deep Learning Models from the Underlying Hardware

  arXiv (Cornell University) · 2020-09-20 · 4 citations

  preprintOpen accessSenior author

  State-of-the-art deep learning systems such as TensorFlow and PyTorch tightly couple the model with the underlying hardware. This coupling requires the user to modify application logic in order to run the same job across a different set of resources, thereby limiting the choice of hardware for a given workload and potentially forcing the user to forgo more efficient hardware configurations. We propose VirtualFlow, a system leveraging a novel abstraction called virtual node processing to decouple the model from the hardware. In each step of training or inference, the batch of input data is split across virtual nodes instead of hardware accelerators (e.g. GPUs and TPUs). Mapping multiple virtual nodes to each accelerator and processing them sequentially effectively time slices the batch, thereby allowing users to reduce the memory requirement of their workloads and mimic large batch sizes on small clusters. Using this technique, VirtualFlow enables many new use cases, such as reproducing training results across different hardware, resource elasticity, and heterogeneous training. In our evaluation, our implementation of VirtualFlow for TensorFlow achieved strong convergence guarantees across different hardware with out-of-the-box hyperparameters, up to 48% lower job completion times with resource elasticity, and up to 42% higher throughput with heterogeneous training.

  PublisherOA PDFDOI

• PrismDB: Read-aware Log-structured Merge Trees for Heterogeneous Storage.

  arXiv (Cornell University) · 2020 · 4 citations

  Senior authorCorresponding
  • Computer Science
  • Computer Science
  • Computer hardware

  In recent years, emerging hardware storage technologies have focused on divergent goals: better performance or lower cost-per-bit of storage. Correspondingly, data systems that employ these new technologies are optimized either to be fast (but expensive) or cheap (but slow). We take a different approach: by combining multiple tiers of fast and low-cost storage technologies within the same system, we can achieve a Pareto-efficient balance between performance and cost-per-bit. This paper presents the design and implementation of PrismDB, a novel log-structured merge tree based key-value store that exploits a full spectrum of heterogeneous storage technologies (from 3D XPoint to QLC NAND). We introduce the notion of read-awareness to log-structured merge trees, which allows hot objects to be pinned to faster storage, achieving better tiering and hot-cold separation of objects. Compared to the standard use of RocksDB on flash in datacenters today, PrismDB's average throughput on heterogeneous storage is 2.3$\times$ faster and its tail latency is more than an order of magnitude better, using hardware than is half the cost.

  Publisher

Recent grants

• CAREER: Towards Scalable Datacenter Services with Strong Robustness Guarantees

  NSF · $529k · 2010–2017

• FIA: Collaborative Research: NEBULA: A Future Internet That Supports Trustworthy Cloud Computing

  NSF · $503k · 2010–2014

• Collaborative Research: NeTS-ANET: A Network Architecture for Federated Virtual/Physical Worlds

  NSF · $509k · 2008–2013

• NeTS: Medium: Collaborative Research: Designing a Content-Aware Internet Ecosystem

  NSF · $340k · 2009–2013

• TC: Large: Collaborative Research: Facilitating Free and Open Access to Information on the Internet

  NSF · $750k · 2012–2018

Frequent coauthors

• David Mazières

  Stanford University

  18 shared

• Jennifer Rexford

  18 shared

• Matvey Arye

  13 shared

• Siddhartha Sen

  11 shared

• Wyatt Lloyd

  11 shared

• Michael Kaminsky

  Carnegie Mellon University

  11 shared

• Erik Nordström

  Skåne University Hospital

  9 shared

• David G. Andersen

  Carnegie Mellon University

  9 shared

Labs

• Center for Information Technology PolicyPI

Awards & honors

• ACM Grace Murray Hopper Award
• ACM SIGOPS Mark Weiser Award
• ACM Fellow
• Presidential Early Career Award for Scientists and Engineers…
• SIGCOMM Test of Time Award