DSA Research Group blog — news, updates, and insights on distributed systems and cloud computing research.
News and technical insights from the Distributed Systems Architecture Research Group.
SOSP 2026 met in Prague September 29 to October 2. A reading list from the accepted papers on verification, LLM serving, shared logs and operations.
Read more →
SafarDB moves replicated-data-type transaction logic onto the FPGA-resident NIC itself, reporting 6.8x latency and throughput gains over RDMA-based replication.
Read more →
gcsfs 2026.8.0 makes concurrent prefetching the default read path, turning cloud storage reads from a network wait into a compute-bound operation.
Read more →
A systematic review of 122 papers on serverless HPC, AI, and big data maps eight years of research trying to fit compute-intensive workloads onto FaaS.
Read more →
A USENIX ATC 2025 paper called Barre shows how three small changes to an old AIMD control loop cut AI-training congestion at 400 Gbps with no new hardware.
Read more →
Kubernetes v1.36 moves gang scheduling into kube-scheduler itself with a decoupled PodGroup API — resolving a decades-old coscheduling problem inside the …
Read more →
OSDI's 20th edition met in Seattle in July 2026. The papers that matter for cloud infrastructure show a field reorganizing itself around LLM serving economics.
Read more →
SoCC's 17th edition meets in Singapore in November 2026. Its call for papers is the clearest available statement of what cloud research now considers in scope.
Read more →
Leaderless and multi-leader consensus is reaching production in Cassandra's Accord and Multi-Raft systems — and it changes what a cloud data API can promise …
Read more →
Tracing the research thread from Globus Toolkit and DRMAA through GridWay's metascheduler model to today's Kubernetes multi-cluster orchestration.
Read more →
Distributed-systems research on scheduling, gang coordination, and resource federation shaped Kubernetes into the substrate for GPU-based AI infrastructure.
Read more →
How WebAssembly's startup speed and WASI standardization are displacing containers for edge workloads — and where the limits lie.
Read more →Sovereign cloud requirements, open infrastructure alternatives to hyperscaler lock-in, and how organizations are building vendor-neutral cloud strategy in 2026.
Read more →Amazon Web Services platform overview — compute, storage, networking, and pricing structure.
Read more →Decision framework for matching workload requirements to cloud provider strengths.
Read more →How AWS, Azure, GCP, and others handle automatic scaling — policies, response times, and cost implications.
Read more →Reserved instances, spot pricing, and cost monitoring tools across providers.
Read more →Managed relational, NoSQL, and in-memory databases across major cloud providers.
Read more →Workload assessment frameworks, migration patterns, and common pitfalls.
Read more →VPC architecture, load balancing, and inter-region connectivity across major providers.
Read more →Compute, storage, and network throughput comparisons across AWS, Azure, GCP, and OCI.
Read more →Identity management, encryption, and compliance certifications across AWS, Azure, and GCP.
Read more →Object storage, block storage, and file storage across AWS S3, Azure Blob, GCS, and others.
Read more →Comparing managed Kubernetes services — EKS, AKS, GKE, and alternatives.
Read more →Simplified cloud infrastructure for startups and developers — droplets, managed databases, and app platform.
Read more →AWS Outposts, Azure Stack, Google Distributed Cloud, and the edge computing landscape.
Read more →GCP strengths in data analytics, machine learning, and Kubernetes-native infrastructure.
Read more →How academic grid computing research led to modern cloud infrastructure.
Read more →Comparing infrastructure, platform, and software as a service models — when to use each and what to expect.
Read more →IBM Cloud's positioning for regulated industries, mainframe modernization, and hybrid cloud with Red Hat.
Read more →Azure platform capabilities, enterprise integration, and how it compares to other major cloud providers.
Read more →When multi-cloud makes sense, vendor lock-in risks, and orchestration tools.
Read more →How OpenNebula provides enterprise cloud management without vendor lock-in.
Read more →OCI's autonomous database, high-performance compute, and pricing advantages for Oracle shops.
Read more →Deployment model comparison covering security, cost, and flexibility trade-offs.
Read more →AWS Lambda vs Azure Functions vs Google Cloud Functions — cold starts, pricing, and runtime support.
Read more →Overview of cloud computing models, deployment types, and core benefits for organizations evaluating infrastructure options.
Read more →How trusted execution environments — NVIDIA confidential GPU modes, Intel TDX, AMD SEV-SNP — protect AI models and data while in use on shared GPU …
Read more →OpenNebula's 2026 release deepens NVIDIA GPU and confidential-computing integration, positioning open private clouds as a credible home for AI workloads outside …
Read more →The economics of placing AI inference at the edge versus in centralized GPU clusters — latency, model size, power, and data gravity, and how to decide where a …
Read more →Banks are piloting hybrid key exchange combining X25519 with ML-KEM in TLS 1.3 for cloud APIs. What hybrid post-quantum TLS is, why finance moved first, and …
Read more →A year-end look at the forces that reshaped cloud infrastructure in 2025 — the AI inference economy, the Data Act, post-quantum cryptography, open-source …
Read more →What the October 2025 AWS and Azure outages reveal about failure domains, control-plane dependencies, and the limits of multi-region resilience.
Read more →How carbon-aware Kubernetes scheduling and cloud emissions reporting intersect with the EU's CSRD Scope 3 requirements, and what infrastructure teams should …
Read more →Two years after Broadcom closed the VMware deal, a look at where private-cloud workloads actually migrated — Proxmox, OpenNebula, Nutanix, public cloud, or …
Read more →The EU Data Act's cloud-switching provisions become applicable on 12 September 2025 — what the new obligations require of providers and how customers should …
Read more →As Kubernetes becomes standard infrastructure in financial services and healthcare, auditors are setting concrete expectations for container security, change …
Read more →Hyperscalers are building European sovereign cloud entities while open-source stacks offer a different sovereignty model — comparing the two approaches as …
Read more →Redis 8 adopts the AGPLv3 open-source license a little over a year after moving to source-available terms — what the reversal means for Valkey and the broader …
Read more →Why rising rack power densities have pushed direct-to-chip and immersion liquid cooling from niche deployment to mainstream necessity in AI datacenters.
Read more →NIST selects HQC as a backup key-encapsulation mechanism built on different mathematical assumptions than ML-KEM — what the fifth post-quantum algorithm means …
Read more →Why electrical capacity, grid interconnection queues, and on-site generation have replaced silicon as the binding constraint on cloud datacenter growth.
Read more →DeepSeek's efficient models prompted a market reassessment of AI infrastructure spending. We examine what it means for the economics of inference at scale.
Read more →A year-end review of 2024 in cloud infrastructure — egress-fee removals, the EU Data Act, the VMware exodus, and the fork playbook all pushed against lock-in.
Read more →A new class of GPU-specialized cloud providers has emerged to serve AI workloads. We examine the neocloud model, what it offers, and the risks it carries.
Read more →The major clouds are all building custom ARM server chips. We compare AWS Graviton4, Azure Cobalt, and Google Axion, and what custom silicon means for …
Read more →NIST has finalized its first post-quantum cryptography standards as FIPS 203, 204, and 205. We lay out what they are and a practical cloud migration timeline.
Read more →A faulty CrowdStrike sensor update crashed millions of Windows machines worldwide. We examine what the outage reveals about single-vendor risk and update …
Read more →Kubernetes turns ten. We trace its path from Google's Borg lineage to becoming the default substrate for running workloads, and what its maturity now demands.
Read more →Redis adopted source-available licensing and the community forked it as Valkey under the Linux Foundation. We trace the now-familiar relicense-and-fork pattern.
Read more →A backdoor was found in the xz-utils compression library after a multi-year social-engineering campaign. We examine what it reveals about open-source trust.
Read more →The EU Data Act has entered into force with cloud-switching provisions on a phased runway. We explain what the Act requires and what cloud customers actually …
Read more →Google Cloud will waive egress fees for customers leaving the platform, citing the EU Data Act. We examine what changed and whether the other hyperscalers will …
Read more →Broadcom has ended VMware perpetual licensing. We survey the realistic migration options for private cloud — Proxmox, OpenNebula, OpenStack, Nutanix, and …
Read more →Broadcom has closed its acquisition of VMware. We examine the pattern from Broadcom's past acquisitions and what virtualization customers should watch for next.
Read more →What a software bill of materials is, why post-Log4Shell policy made SBOMs a requirement, and how SLSA and SBOM tooling fit into cloud supply-chain security.
Read more →HashiCorp moved Terraform to the Business Source License. We trace what changed, why the community forked, and what OpenTofu means for infrastructure-as-code …
Read more →37signals moved its workloads off the public cloud and published the numbers. We examine what cloud repatriation actually involves and which workloads it suits.
Read more →