AI Metrics with Grafana Cloud

The Grafana AI Metrics dashboard shown above tracks performance metrics for AI/ML RoCEv2 network traffic, for example, large scale CUDA compute tasks using NVIDIA Collective Communication Library (NCCL) operations for inter-GPU communications: AllReduce, Broadcast, Reduce, AllGather, and ReduceScatter.

The metrics include:

  • Total Traffic Total traffic entering fabric
  • Operations Total RoCEv2 operations broken out by type
  • Core Link Traffic Histogram of load on fabric links
  • Edge Link Traffic Histogram of load on access ports
  • RDMA Operations Total RDMA operations
  • RDMA Bytes Average RDMA operation size
  • Credits Average number of credits in RoCEv2 acknowledgements
  • Period Detected period of compute / exchange activity on fabric (in this case just over 0.5 seconds)
  • Congestion Total ECN / CNP congestion messages
  • Errors Total ingress / egress errors
  • Discards Total ingress / egress discards
  • Drop Reasons Packet drop reasons
AI Metrics with Prometheus and Grafana describes how to stand up an analytics stack with Prometheus and Grafana to track performance metrics for an AI/ML GPU cluster. This article shows how to integrate with Prometheus and Grafana hosted in the cloud, Grafana Cloud, instead of running the services locally.

Note: Grafana Cloud has a free service tier that can be used to test this example.

Continue reading

NB530: Broadcom Hits 102.4 Tbps With Tomahawk 6; Wireshark Debuts Certificate Program

Take a Network Break! We start with two critical vulnerabilities: one affecting cloud versions of Cisco ISE, and the other for HPE StoreOnce. In the news, Broadcom announces the Tomahawk 6 ASIC with 102.4Tbits of bandwidth, SentinelOne suffers a self-imposed network outage, and the Wireshark Foundation announces its first-ever professional certification for Wireshark. Cisco rebrands... Read more »

Worth Reading 060925


The story of computing and communications over the past eighty years has been a story of quite astounding improvements in the capability, cost and efficiency of computers and communications.


In recent discussions, it became clear that additional information could be helpful, breaking down what a user or administrator needs to understand about TLS implementation and configuration options to better assess points of potential exposure.


The use of pseudo-random processes to generate secret quantities can result in pseudo-security. A sophisticated attacker may find it easier to reproduce the environment that produced the secret quantities and to search the resulting small set of possibilities than to locate the quantities in the whole of the potential number space.


We’ve all had the serendipity experience, even online — clicking through a chain of links, scanning Google search results, drifting between loosely connected ideas. But search engines and information retrieval systems aren’t designed to enhance serendipity.


Here I want to look at just one day of the operation of the Internet’s BGP network by looking at the behaviour of a single BGP session. The day we’ll use for this study is the 8h May 2025, and the BGP vantage point used here is an unremarkable Continue reading

Getting Started with the Pytest Plugin for Infrahub

Getting Started with the Pytest Plugin for Infrahub

We all write code, but how do we know the changes we make in the future won’t break something that used to work? That’s where testing becomes important.

The idea is to catch problems early, ideally before they reach production. In the Python world, one of the most common ways to do this is with a tool called pytest. It lets you write tests to check that your code behaves the way you expect and helps you catch issues before they become a bigger problem.

Originally published under - https://www.opsmill.com/pytest-plugin-infrahub/

When working with Infrahub, testing is just as important. You might want to make sure your GraphQL queries are valid, your Jinja2 templates render correctly, or your transformations behave as expected.

Infrahub simplifies this by offering a pytest plugin that doesn’t require Python code; you define tests using plain YAML. This makes testing more accessible to teams across roles and speeds up the feedback loop during development.

These kinds of unit tests aren’t just about convenience, they help establish a production-ready automation system. With automated checks built into your process, every change is validated consistently, reducing the chance of something breaking unexpectedly. That consistency builds trust when your Continue reading

Publishing Content as an Introvert

I got an interesting question from a reader. He listened to my podcast with Eric Chou and decided to try to learn in public:

Currently, I’m studying for the CCNP ENARSI exam, and would like to start posting my labs to LinkedIn, and perhaps even upload my lab topologies and configs to Git.

That’s a great idea. I would minimize the LinkedIn part1 and focus on Git:

Read more …

Deploying ELK Stack with Docker Compose (2025 Edition)

This guide walks you through installing and configuring the ELK Stack (Elasticsearch, Logstash, Kibana, and Filebeat) using Docker Compose. It is fully updated for Elasticsearch 9.0.2 and explains the necessary changes for versions 8+ and above, including the required security setup and user permissions. Prerequisites Ensure you have the following installed on your system: The […]

<p>The post Deploying ELK Stack with Docker Compose (2025 Edition) first appeared on IPNET.</p>

UGreen NASync DXP2800 Review and First Impressions

UGreen NASync DXP2800 Review and First Impressions

TL;DR - For anyone who doesn’t want to go through the full post, here’s the short version. I bought the UGreen NASync DXP2800 (2 bay) from Amazon for £249 and paired it with two Seagate Ironwolf 8TB HDDs, around £180 each.

The unit comes with an Intel N100 CPU, 8GB of RAM (upgradeable to 16GB, but there’s only one RAM slot), and a 2.5Gb/s LAN port. It has a solid build, was easy to set up, and I actually like the UI. Sure, it lacks a lot of features compared to Synology or QNAP, but since I’m mainly using it for file storage, I’m happy with the purchase.

Setting up Proxmox Backup Server
In this post, we’ll go through the process of setting up Proxmox Backup Server and backing up all the VMs from my Proxmox server to this backup server. So, let’s get to it.
UGreen NASync DXP2800 Review and First ImpressionsPacketswitchSuresh Vina
UGreen NASync DXP2800 Review and First Impressions

But Why UGreen NAS?

The short answer is, this is the best bang for the buck. For £249, I’m getting a 2-bay NAS with an N100 CPU, 8GB of RAM, a 2.5Gb/s LAN port, and two NVMe slots.

I’ve been wanting to buy a NAS for over Continue reading

HN784: Accelerate Your Network Automation With Gluware Labs and New Ansible Collection (Sponsored)

Network automation is today’s topic with sponsor Gluware. Gluware provides a network automation platform that targets both network engineers and automation builders. On today’s Heavy Networking, we discuss how Gluware supports these two constituencies. We also talk about a recent product announcement, Gluware Labs. Gluware Labs includes a free Community Edition of Gluware software you... Read more »

Dell’s Advice To Enterprises: Buy AI, Don’t Try To Build It

Unsurprisingly, the main topic of conversation at the recent Dell Technologies World 2025 event in Las Vegas was AI, and a central theme that wove through many of the messages we heard there was that adopting the emerging technology is much easier now than it was even a year ago. …

Dell’s Advice To Enterprises: Buy AI, Don’t Try To Build It was written by Jeffrey Burt at The Next Platform.

Finding End-to-End Paths: Topology and Endpoints

We know there are three main ways to move packets across a network. However, before we can start forwarding packets, someone has to populate the forwarding tables in the intermediate devices or build the sequence of nodes to traverse in source routing.

Usually, whoever is responsible for the contents of the forwarding tables must first discover the network topology. Let’s start there, using the following network diagram to illustrate the discussion.

Read more …

Juniper vJunos-router in Containerlab

Juniper vJunos-router in Containerlab

If you follow me or read my blog, you probably know I'm a big advocate of Containerlab. I've been using it for over two years now and I absolutely love it. Why? Because it's open source, it has an amazing community behind it (thank you again, Roman), and labs are defined using simple YAML files that are easy to share and reuse.

So far, I've used Cisco IOL, Arista EOS, and Palo Alto VM in Containerlab. And finally, the time came to try Juniper. I decided to test the Juniper vJunos-router, which is a virtualized MX router. It's a single-VM version of vMX that doesn't require any feature licenses and is meant for lab or testing purposes. You can even download the image directly from Juniper's website without needing an account. Thank you, Juniper and Cisco, please take note. In this post, I'll show you how to run Juniper vJunos-router in Containerlab.

Prerequisites

This post assumes you're somewhat familiar with Containerlab and already have it installed. If you're new, feel free to check out my introductory blog below. Containerlab also has great documentation on how to use vJunos-router, so be sure to check that out as well.

Containerlab - Creating Continue reading

N4N029: Four Goals of Network Design

Let’s explore four goals of network design: stability, speed, scalability, and security. These goals are based on Ethan’s experience designing, building, and operating networks. Network architects and design experts might have other objectives, and that’s fine, but these four goals are the basis of today’s episode. Ethan and Holly discuss why these four goals are... Read more »

Weird: Ports on Linux Bridge Are Stuck

Just when you thought you got used to the weirdnesses in the networking implementations, you get a curveball like this one. Life is never dull if you test network devices.

Before releasing netlab release 2.0, I ran the full suite of integration tests for all devices for which I have the images. Interestingly, most VXLAN tests failed for Cumulus Linux 4.x even though we haven’t touched that code for ages.

Next step: trying to figure out what changed. The configuration changes were minimal. Even worse, the failure was non-deterministic. Somehow, we managed to transform a Cumulus Linux 4.x VM into a Heisenberg switch.

Read more …

D2DO274: Firefly’s State of IaC Report for 2025, aka ClickOps Is a Disgrace (Sponsored)

Firefly is a cloud infrastructure automation platform that helps cloud teams, DevOps, SRE, platform engineering, DevSecOps, and other groups manage their entire cloud as code. Firefly helps to manage cloud complexity and produce consistent and efficient cloud platforms with code. To help Firefly better understand their customers and industry trends around Infrastructure as Code (IaC),... Read more »

HPE Uses AI To Drive The Business, Which Is Increasingly AI

Hewlett Packard Enterprise is going through yet another restructuring to reduce costs, something we have seen a lot of in the past two decades and a half decades since it acquired Compaq to become a volume server peddler as well as high end system supplier for enterprises. …

HPE Uses AI To Drive The Business, Which Is Increasingly AI was written by Timothy Prickett Morgan at The Next Platform.

1 … 40 41 42 43 44 … 3,833