Talk Title: Timeless and Timely Observations About Asynchrony

Abstract: This talk will outline a "Delaying the Future" theorem for asynchronous systems. It captures the essential role that the happens-before relation plays in knowledge, coordination and linearizability in asynchronous computation. We will survey both classic and more recent results. The latter will be based in large part on joint work with Raissa Nataf.

Bio: Yoram Moses completed his BSc at the Hebrew University in 1981, a PhD at Stanford in 1986 which was followed by a post-doc at MIT. His research involves theoretical aspects of distributed and multi-agent systems. He is a co-author of the book Reasoning about Knowledge and received the Goedel and Dijkstra prizes for initiating work on knowledge in distributed systems.

Talk title: Byzantine Consensus: Modern Problem Requires Traditional Solution

Abstract: Byzantine consensus is a central problem in distributed computing. The problem comes in multiple flavors -- for example, depending on synchrony, validity, and recently, the availability of AI predictions. The myriad settings seem to imply we need different approaches for each if we want to optimize performance. But what if I tell you that one general approach is good enough? This talk will go through a common and traditional framework for Byzantine consensus, in particular, one which relies on a weaker primitive called graded consensus AKA adopt-commit to guard "safety" of the protocol. We will see how different instantiations of this framework achieve new results, solving long-standing open problems and addressing new settings.

Bio: Muhammad Ayaz Dzulfikar is a final year PhD student in National University of Singapore, advised by Seth Gilbert. His research interests are in distributed algorithms, with focus on Byzantine consensus.

Talk title: Rethinking Safe Memory Reclamation for Concurrent Data Structures

Abstract: Safe memory reclamation is crucial to memory safety for optimistic and non blocking concurrent data structures in non garbage collected programming languages.

Yet, designing an ideal SMR algorithm remains challenging. The main difficulties include achieving high speed and scalability, keeping the interface easy to use for programmers, ensuring applicability to a broad class of data structures, bounding the memory footprint, and avoiding asymmetric overhead across data structure operations. Addressing these issues is critical for enabling the deployment of concurrent data structures in production systems.

In this talk, I will present three new methods for safe memory reclamation that are each designed to address a distinct shortcoming of state-of-the-art safe memory reclamation algorithms. The first method tackles the tradeoffs among key properties—speed, bounded memory footprint, usability and wide applicability—by delivering fast, memory-efficient, and broadly applicable reclamation without intrusive changes or specialized hardware support. The second method reduces the uneven overheads observed in techniques such as Hazard Pointers, where reclamation steps impose disproportionate overhead on data structure operations. The third method rethinks the delayed reclamation paradigm of designing safe memory reclamation algorithms by enabling immediate reuse of memory, achieving the ideal memory footprint of sequential data structures while preserving the performance benefits of concurrent ones.

Together, these methods demonstrate how codesigning safe memory reclamation algorithms with neighboring layers of the programming environment, including operating system and hardware cache-coherence, can overcome longstanding challenges efficiently.

I will close with a broader question: co-design with systems and hardware enabled me to develop new approaches to safe memory reclamation—what concurrency problems might such co-designs help you solve?

Bio: Ajay Singh obtained his PhD in Computer Science from University of Waterloo, advised by Trevor Brown and Petr Buhr, with research centered around designing safe memory reclamation algorithms for concurrent data structures. He is a recipient of the Distinguished Paper Award at PPoPP 2025, the Best Artefact Award at PPoPP 2021, the David R Cheriton Graduate Scholarship (awarded to top graduate students of the department) 2022.

Currently, he is a postdoctoral researcher at Institute of Computer Science (ICS) FORTH in Greece, working with Panagiota Fatourou. His research revolves around understanding challenges in designing and deploying concurrent data structures and algorithms on current and emerging computing platforms, drawing on insights from subfields such as Computer Architecture, Programming Languages, Operating Systems, and Networks, and in turn enriching these areas through novel solutions.

Talk title: Accord: Optimal Leaderless Replication in Apache Cassandra

Abstract: Classical leader-driven consensus protocols like Paxos form the foundation of state-machine replication (SMR) but their reliance on a leader creates a performance bottleneck. Over the years, multiple protocols have attempted to improve on Paxos by removing the leader from the critical path, but each introduced new challenges such as suboptimal fault tolerance, correctness issues, or high tail latency. Accord is a new leaderless protocol that addresses these problems. It eliminates the single point of coordination, is fast under favorable conditions, and we have proven it tolerates the optimal number of failures. Already merged into Apache Cassandra and being deployed at Apple, Accord is the first major leaderless SMR protocol adopted in industry. This talk will trace the path from Paxos to Accord, highlighting how Accord overcomes the limitations of previous protocols to enable practical, scalable SMR.

Bio: Fedor Ryabinin is a PhD student at the IMDEA Software Institute advised by Alexey Gotsman. His research focuses on distributed computing, particularly consensus and state-machine replication protocols.

Talk title: Alpenglow: Solana's new consensus protocol

Abstract: Solana has risen to prominence as a practical high-throughput, low-latency blockchain. However, the platform has faced its fair share of struggles along the way. Recently, a far-reaching consensus overhaul was proposed and accepted by Solana's community. This talk explains the protocol — Alpenglow — and its contribution to blockchain research. https://www.solanaroadmap.com/

Bio: Kobi is a blockchain researcher working on Solana's consensus. He studied at Oxford, did his PhD at ETH Zurich, worked in the blockchain domain at Dfinity and now Anza.