AI workloads are shifting towards smaller transfers where latency is crucial.
“Homa...can reduce tail latency by an order of magnitude or more.”
networking
11 tracked signals on networking.
Meta open-sourced MetaRoCE, a new RDMA transport protocol built for million-GPU AI clusters on Ethernet.
“The fabric sees packets, but the NIC sees intent.”
NVIDIA Spectrum-X Ethernet reengineers networking for giga-scale AI GPU clusters
NVIDIA BlueField-4 DPU enables dedicated networking for agentic AI factory infrastructure
“Agentic AI factories connect diverse users, agents, applications, data sources, and storage systems to massively accelerated compute at multi-terabit bandwidth per server, making dedicated DPU processing essential for line-rate networking, storage, and security.”
Databricks serverless platform launches tens of millions of VMs daily at massive scale
NVIDIA Quantum InfiniBand adds one-click intent-based multi-tenant security profiles to Unified Fabric Manager.
“This cuts deployment time to minutes from hours or days”
Databricks Inbound Private Link now covers account-level Genie One, console, and custom URLs
Meta open-sources authenticated NTP via NTS at nts.meta.com, making time verifiable
“That work made time precise. This work makes it verifiable.”
AWS VPC Lattice product manager discusses the application networking service's roadmap and operations.
AWS engineers demo a methodology to trace surprise data transfer charges using VPC Flow Logs and DNS query correlation.
“you get an AWS bill and you look at your data transfer charges and you're like, "Why is it so high?”
An AWS networking team livestream discusses IPv6 adoption and VPC Lattice basics.
“I really really love IPv6 and I think everyone should be doing IPv6.”