分布式、并行与集群计算

We design and implement parallel prefix sum (scan) algorithms using Ascend AI accelerators. Ascend accelerators feature specialized computing units: the cube units for efficient matrix multiplication and the vector units for optimized…

This work addresses the inherent issues of high latency in blockchains and low scalability in traditional consensus protocols. We present pod, a novel notion of consensus whose first priority is to achieve the physically-optimal latency of…

Sensing systems powered by energy harvesting have traditionally been designed to tolerate long periods without energy. As the Internet of Things (IoT) evolves towards a more transient and opportunistic execution paradigm, reducing energy…

Modern ML training and inference now span tens to tens of thousands of GPUs, where network faults can waste 10--15\% of GPU hours due to slow recovery. Common network errors and link fluctuations trigger timeouts that often terminate entire…

Modern cloud-native systems increasingly rely on multi-cluster deployments to support scalability, resilience, and geographic distribution. However, existing resource management approaches remain largely reactive and cluster-centric,…

With the development of large-scale models, traditional distributed bilevel optimization algorithms cannot be applied directly in low-resource clients. The key reason lies in the excessive computation involved in optimizing both the lower-…

As LLMs and foundation models scale, checkpoint/restore has become a critical pattern for training and inference. With 3D parallelism (tensor, pipeline, data), checkpointing involves many processes, each managing numerous tensors of varying…

Federated Learning (FL) enables mobile edge devices, functioning as clients, to collaboratively train a decentralized model while ensuring local data privacy. However, the efficiency of FL in wireless networks is limited not only by…

Edge computing enables latency-critical applications to process data close to end devices, yet task heterogeneity and limited resources pose significant challenges to efficient orchestration. This paper presents a measurement-driven,…

Cloud data pipelines increasingly operate under dynamic workloads, evolving schemas, cost constraints, and strict governance requirements. Despite advances in cloud-native orchestration frameworks, most production pipelines rely on static…

We initiate the study of approximate maximum matching in the vertex partition model, for graphs subject to dynamic changes. We assume that the $n$ vertices of the graph are partitioned among $k$ players, who execute a distributed algorithm…

We initiate the study of deterministic distributed graph algorithms with predictions in synchronous message passing systems. The process at each node in the graph is given a prediction, which is some extra information about the problem…

Because strongly-linearizable objects provide stronger guarantees than linearizability, they serve as valuable building blocks for the design of concurrent data structures. Yet, many objects that have linearizable implementations from base…

Distributed filesystem metadata updates are typically synchronous. This creates inherent challenges for access efficiency, load balancing, and directory contention, especially under dynamic and skewed workloads. This paper argues that…

The NOSTR is a communication protocol for the social web, based on the w3c websockets standard. Although it is still in its infancy, it is well known as a social media protocol, with thousands of trusted users and multiple user interfaces,…

The Adaptable TeaStore specification provides a microservice-based case study for implementing self-adaptation through a control loop. We argue that implementations of this specification should be informed by key properties of…

This paper presents how an existing framework for offline performance optimization can be applied to microservice applications during the Release phase of the DevOps life cycle. Optimization of resource allocation configuration parameters…

Consistent hashing is fundamental to distributed systems, but ring-based schemes can exhibit high peak-to-average load ratios unless they use many virtual nodes, while multi-probe methods improve balance at the cost of scattered memory…

The proliferation of Large Language Models (LLMs) has been accompanied by a reliance on cloud-based, proprietary systems, raising significant concerns regarding data privacy, operational sovereignty, and escalating costs. This paper…

Large Language Models (LLMs) are rapidly being integrated into real-world applications, yet their autoregressive architectures introduce significant inference time variability, especially when deployed across heterogeneous edge-cloud…