GPU
Sep 16, 2026·9 min read
ECMP Hash Collisions in a Fat-Tree: Too Few Flows, Not Too Many
A k=8 fat-tree has 16 equal-cost paths per pod pair, so why does AllReduce run at 60% of line rate? ECMP fails on AI traffic because collectives produ...
BaaZ Blogs
Technical deep-dives on GPU infrastructure, distributed training, and cloud-native systems.
GPU
Sep 16, 2026·7 min read
DAC, ACC or AOC: Cabling GPU Nodes to the Switch in the Same Rack
Passive DAC is the right default for GPU-to-switch links inside a rack - but at NDR 400/800G its reach stops at 1.5 m. How to split DAC, active copper...
GPU
Sep 8, 2026·5 min read
Contributing a Queue Validator to the KAI Scheduler
How we contributed a validating admission webhook to NVIDIA's KAI Scheduler that enforces parent-child queue quota consistency - preventing resource o...
GPU
Sep 8, 2026·13 min read
Dual-Network Kubernetes Pods with RDMA on NVIDIA GH200: A Hands-On Guide
How to set up a management network and a dedicated RDMA training network inside Kubernetes pods - from hardware discovery to working RDMA verbs, on a ...
GPU
Sep 8, 2026·9 min read
How to Calculate if Your Network is Bottlenecking Distributed Training
A practical guide to understanding why your multi-node GPU training might be slower than expected.
GPU
Apr 13, 2026·14 min read
Understanding Secondary Networks for GPU Workloads in Kubernetes
Macvlan, host-device, SR-IOV, and IPoIB - what they are, how they differ, and when to use each for RDMA and NCCL traffic in GPU training clusters.
GPU
Apr 6, 2026·6 min read
Adding Global Config to the NVIDIA Network Operator
Global config for NicClusterPolicy: set repository, version, and imagePullSecrets once instead of repeating across every Network Operator component.
GPU
Jan 18, 2026·5 min read
GPU-to-GPU Communication Across Nodes: What Actually Works
A practical guide to NVLink, NVSwitch, InfiniBand, RoCE, and GPUDirect for multi-node GPU clusters. Cut through the jargon and understand what hardwar...
GPU
Aug 27, 2025·5 min read
Understanding Power Management in GPU via PCIe
Learn how modern GPUs implement intelligent power management through PCIe generation scaling and why your high-end GPUs might be operating at Gen 1 sp...
Sep 8, 2026·6 min read
Centralised Control Planes for SaaS - Part 1: SaaS Business Models
BaaZ now specializes in GPU infrastructure consulting for AI. See our GPU infrastructure services →
Sep 8, 2026·7 min read
Centralised Control Planes for SaaS - Part 2: Stateless Async Event Handling
BaaZ now specializes in GPU infrastructure consulting for AI. See our GPU infrastructure services →
Sep 8, 2026·2 min read
Centralised Control Planes for SaaS - Part 3: Logical and Physical Models
BaaZ now specializes in GPU infrastructure consulting for AI. See our GPU infrastructure services →
Sep 8, 2026·7 min read
State Management for Infra Products - Part 1: State Machines vs Observed State
BaaZ now specializes in GPU infrastructure consulting for AI. See our GPU infrastructure services →
Oct 2, 2025·7 min read
Understanding RX vs TX - Making Sense of Network Traffic Direction with Real Examples
Learn the fundamentals of RX (receive) and TX (transmit) metrics in network monitoring through real-world Mellanox ONYX switch examples, common traffi...
May 17, 2023·2 min read
Demystifying Helm and Operator pattern.
Discover insights and explanations about Helm and the Operator Pattern. Gain a clearer understanding of these tools and patterns used for efficient Ku...