12 papers selected.
FaaSCtrl: A Comprehensive-Latency Controller for Serverless Platforms
Abhisek Panda, Smruti R. Sarangi
TL;DR — FaaSCtrl is a feedback-control system for serverless platforms that jointly manages cold-start, queuing, and execution latency to meet end-to-end SLOs.
Why notable — One of the few serverless controllers that addresses all three latency components together, providing a principled alternative to ad-hoc autoscaling heuristics.
FUSIONIZE++: Improving Serverless Application Performance Using Dynamic Task Inlining and Infrastructure Optimization
Trever Schirmer, Joel Scheuner, Tobias Pfandzelter, David Bermbach
TL;DR — FUSIONIZE++ dynamically fuses serverless functions at runtime and co-optimizes infrastructure selection to cut invocation overhead and cost.
Why notable — Demonstrates concrete end-to-end performance gains from function fusion in real serverless deployments, directly relevant to practitioners optimizing FaaS pipelines.
BaaSLess: Backend-as-a-Service (BaaS)-Enabled Workflows in Federated Serverless Infrastructures
Thomas Larcher, Philipp Gritsch, Stefan Nastic, Sashko Ristov
TL;DR — BaaSLess integrates BaaS capabilities into serverless workflow execution across federated multi-provider infrastructures, enabling stateful cross-cloud function orchestration.
Why notable — Addresses the underexplored intersection of BaaS, serverless, and federation, offering a practical blueprint for multi-cloud serverless applications.
Slim and Fast: Low-Overhead Container Overlay Network With Fast Connection Setup
Fusheng Lin, Xin Zhang 0117, Guo Chen 0001, Li Chen 0008 et al.
TL;DR — A redesigned container overlay network that minimizes control-plane overhead and dramatically reduces connection setup latency for microservice-dense deployments.
Why notable — The implementation targets real Kubernetes environments and directly improves east-west latency for microservices at scale.
Trustless Collaborative Cloud Federation
Bishakh Chandra Ghosh, Sandip Chakraborty 0001
TL;DR — A blockchain-backed protocol that enables resource sharing across competing cloud providers without requiring a trusted third party.
Why notable — Tackles the fundamental trust barrier in multi-cloud federation with a practical, decentralized design that avoids provider lock-in.
An Adaptive Cloud Resource Quota Scheme Based on Dynamic Portraits and Task-Resource Matching
Zuodong Jin, Dan Tao, Peng Qi 0006, Ruipeng Gao
TL;DR — Builds dynamic workload portraits per tenant and matches them to resource quotas in real time, improving utilization while respecting SLAs in IaaS/PaaS clouds.
Why notable — Moves beyond static quota assignment with a data-driven approach validated on production cloud workload traces.
Aggregate Monitoring for Geo-Distributed Kubernetes Cluster Federations
Chih-Kai Huang 0001, Guillaume Pierre
TL;DR — Proposes a scalable monitoring architecture for Kubernetes federations that aggregates metrics across geo-distributed clusters with low overhead.
Why notable — Practical multi-cloud observability is rarely addressed at the federation layer; this work provides a deployable solution with measured performance on real clusters.
Root Cause Analysis for Cloud-Native Applications
Bartosz Zurkowski, Krzysztof Zielinski
TL;DR — A graph-based RCA framework for microservice architectures that correlates traces, metrics, and logs to pinpoint fault origins in cloud-native deployments.
Why notable — Directly applicable to production microservice operations, offering automated diagnosis that reduces mean time to recovery in complex service graphs.
RAM: A Resource-Aware DDoS Attack Mitigation Framework in Clouds
Fangyuan Xing, Fei Tong 0001, Jialong Yang, Guang Cheng 0001 et al.
TL;DR — RAM dynamically allocates cloud resources for DDoS mitigation based on attack intensity, balancing protection effectiveness against resource cost.
Why notable — Combines attack detection and elastic resource provisioning in a single framework, making it immediately relevant for cloud security operations.
Enabling Multi-Layer Threat Analysis in Dynamic Cloud Environments
Salman Manzoor, Antonios Gouglidis, Matthew Bradbury, Neeraj Suri
TL;DR — A multi-layer threat analysis system that correlates security events across IaaS, PaaS, and application layers to detect composite attacks in dynamic cloud deployments.
Why notable — Addresses the gap between per-layer security tools and cross-layer attack detection, which is critical for securing modern cloud stacks.
Hyperion: Hardware-Based High-Performance and Secure System for Container Networks
Myoungsung You, Minjae Seo, Jaehan Kim, Seungwon Shin 0001 et al.
TL;DR — Hyperion offloads container network security enforcement to programmable hardware, achieving line-rate packet processing with strong isolation guarantees.
Why notable — Shows that hardware offload can simultaneously improve both throughput and security in container networking, with real implementation results.
D-STACK: High Throughput DNN Inference by Effective Multiplexing and Spatio-Temporal Scheduling of GPUs
Aditya Dhakal, Sameer G. Kulkarni, K. K. Ramakrishnan
TL;DR — D-STACK multiplexes multiple DNN inference jobs on shared GPUs via spatio-temporal scheduling, significantly increasing throughput without latency SLO violations.
Why notable — Addresses a critical cloud resource management challenge for AI inference services, with a real system implementation and evaluation against production workloads.