12 papers selected.
Recipe: Hardware-Accelerated Replication Protocols: Rethinking Crash Fault Tolerance Protocols for Untrusted Cloud Environments
Dimitra Giantsidi, Emmanouil Giortamis, Julian Pritzi, Maurice Bailleu et al.
TL;DR — Redesigns crash fault-tolerance protocols using hardware acceleration (TEEs/SmartNICs) to deliver replication with strong guarantees in untrusted cloud environments.
Efficient Performance Guarantees for Function-as-a-Service with Cloud Allocators
Hai Duc Nguyen 0005, Andrew A. Chien
TL;DR — Introduces cloud-allocator abstractions that provide formal performance guarantees for serverless functions, addressing the unpredictability of shared FaaS infrastructure.
Capybara: an Edge-Friendly Distributed Object Store for Diverse Serverless Functions
Xin Chen 0084, Manoj Prabhakar Paidiparthy, Chen Qian 0001, Liting Hu
TL;DR — Proposes a distributed object store purpose-built for edge serverless workloads, enabling diverse function runtimes to share state efficiently at the edge.
EdgeConnector: Enabling Seamless and Efficient Cross-Cluster Device Access in Edge Environment
Yunna Cui, Liwei Shen, Bingkun Sun, Wente Lu et al.
TL;DR — Presents a middleware layer that transparently bridges device access across Kubernetes clusters in heterogeneous edge deployments.
FaaSImage: An Efficient Image Manager for FaaS
Abhisek Panda, Smruti R. Sarangi
TL;DR — Tackles cold-start latency in production FaaS platforms by redesigning container image management with fine-grained layer sharing and prefetching.
Roadrunner: Accelerating Data Delivery to WebAssembly-Based Serverless Functions
Cynthia Marcelino, Thomas W. Pusztai, Stefan Nastic
TL;DR — Designs a high-throughput data-plane runtime that removes I/O bottlenecks for Wasm serverless functions through zero-copy data paths.
Mocha: Scalable and Compliant Function Scheduling for Federated Serverless Computing
Yuqiu Zhang, Hans-Arno Jacobsen
TL;DR — Introduces a scheduler for federated serverless environments that satisfies data-residency and compliance constraints while maintaining high resource utilization.
A Hybrid Runtime for Function-as-a-Service at the Edge
Adam Hall, Umakishore Ramachandran
TL;DR — Combines container and unikernel execution models in a single FaaS runtime to balance isolation, startup latency, and resource efficiency at edge nodes.
Tiaccoon: Unified Access Control with Multiple Transports in Container Networks
Hiroya Onoe, Daisuke Kotani, Yasuo Okabe
TL;DR — Provides a service-mesh-style unified access-control plane that works across heterogeneous transport protocols within container network environments.
MiAR-BFT: Efficient Leaderless Consensus Based on Multi-instance Asynchronous Running for Blockchain
Zhenyu Zhang, Xing Tong, Zhao Zhang 0009, Cheqing Jin
TL;DR — Proposes a leaderless BFT consensus protocol that runs multiple instances concurrently to improve throughput and reduce latency in blockchain middleware.
ER-π: Exhaustive Interleaving Replay for Testing Replicated Data Library Integration
Provakar Mondal, Eli Tilevich
TL;DR — Systematically explores all message interleavings when testing replicated data libraries, surfacing integration bugs that random testing misses in distributed coordination code.
Adjusted Objects: An Efficient and Principled Approach to Scalable Programming
Boubacar Kane, Pierre Sutra
TL;DR — Introduces adjusted objects as a programming abstraction that reconciles strong consistency with scalability, offering a practical alternative to CRDTs for distributed middleware.