The New Era of AI Centers

The New Era of AI Centers

In 1984, Sun was famous for declaring, “The Network is the Computer.” Forty years later we are seeing this cycle come true again with the advent of AI. The collective nature of AI training models relies on a lossless, highly-available network to seamlessly connect every GPU in the cluster to one another and enable peak performance. Networks also connect trained AI models to end users and other systems in the data center such as storage, allowing the system to become more than the sum of its parts. As a result, data centers are evolving into new AI Centers where the networks become the epicenter of AI management.

Container Security: Protect your data with Calico Egress Access Controls

23andMe is a popular genetics testing company, which was valued at $6B in 2021. Unfortunately, there was a massive data breach in December 2023, which caused a steep decline in the company’s value and trust, plummeting the company to a penny stock. While this breach was not directly related to Kubernetes, the same risks apply to containers running in your Kubernetes environments. If your containerized applications do not have the right egress access controls defined, chances of data exfiltration are much higher.

The basics

A typical modus operandi for threat actors is to look for vulnerabilities or misconfiguration in the environment and workloads, install malicious pods through privilege escalation techniques, and then exploit this unsecured pod to exfiltrate data.

Fig 1: Anatomy of a data exfiltration attack
Fig 1: Anatomy of a data exfiltration attack

An easy reconnaissance technique by just scanning the cluster network for public-facing workloads will be a first starting point for most attackers. Privilege escalation occurs mostly due to inconsistent or incorrect RBAC policies in Kubernetes through which unauthorized users can gain root privileges. Vulnerabilities in container images as part of the supply chain are also another attack path. All of these techniques will ultimately land on an exposed pod with a remote code Continue reading

Heavy Strategy: Failure and Resilience

Welcome to a crossover episode with the Heavy Strategy podcast! Firing the wrong person, mistakenly rebooting core switches in a massive network, not passing the CCIE exam– today we talk all about failure. For this conversation, we’re joined by fellow Packet Pushers Kyler Middleton and Ned Bellavance, hosts of the Day Two Cloud podcast. We... Read more »

HS073: Failure and Resilience

Firing the wrong person, mistakenly rebooting core switches in a massive network, not passing the CCIE exam– today we talk all about failure. For this conversation, we’re joined by fellow Packet Pushers Kyler Middleton and Ned Bellavance, hosts of the Day Two Cloud podcast. We swap stories, discuss response and prevention, and talk about accountability,... Read more »

BGP Route Reflectors Considered Harmful

The recent IBGP Full Mesh Between EVPN Leaf Switches blog post generated an interesting discussion on LinkedIn focused on whether we need route reflectors (in small fabrics) and whether they do more harm than good. Here are some of the highlights of that discussion, together with a running commentary.

Please note that we’re talking about BGP route reflectors in reasonably small data center fabrics. Large service provider networks with millions of customer VPN routes are a completely different story. As always, what you read in a random blog post might not apply to your network design. YMMV.
Read more …

Calling Time on DNSSEC

Through the lack of clear signals of general adoption of DNSSEC over three decades, then is it time to acknowledge that DNSSEC is just not going anywhere? Is it time to call it a day for DNSSEC and just move on?

My Network Lab is a Text File

My Network Lab is a Text File

Yes, you read that right. My Network Lab is indeed a text file (YAML file to be more specific). I can share the file with anyone, put it into version control, and never worry about re-creating the lab manually. No more clicking through the GUI and connecting interfaces. How is that even possible? You must be thinking this is clickbait right? Well, I'm talking about using Containerlab to create and manage your network topologies and labs.

My Life Before Containerlab

I started my networking journey with Packet Tracer, then moved on to GNS3. Most of the time, I've used EVE-NG and some Cisco CML. EVE-NG is a great tool, and I still use it for building complex, large topologies with Cisco ISE, multiple firewalls, Active Directory, etc. But when it comes to labbing up pure networking protocols like BGP, OSPF, STP, or even simple IP routing, I needed something very simple that is easy to deploy and manage.

That's when I came across Containerlab which is a Lab-as-a-code tool that helps you set up and manage your network labs easily. Instead of dealing with complex setups and configurations, containerlab simplifies everything for you. Containerlab provides a command-line interface (CLI) that Continue reading

Case Study: NAT64

Introduction

Front

IPng’s network is built up in two main layers, (1) an MPLS transport layer, which is disconnected from the Internet, and (2) a VPP overlay, which carries the Internet. I created a BGP Free core transport network, which uses MPLS switches from a company called Centec. These switches offer IPv4, IPv6, VxLAN, GENEVE and GRE all in silicon, are very cheap on power and relatively affordable per port.

Centec switches allow for a modest but not huge amount of routes in the hardware forwarding tables. I loadtested them in [a previous article] at line rate (well, at least 8x10G at 64b packets and around 110Mpps), and they forward IPv4, IPv6 and MPLS traffic effortlessly, at 45 watts.

I wrote more about the Centec switches in [my review] of them back in 2022.

IPng Site Local

IPng SL

I leverage this internal transport network for more than just MPLS. The transport switches are perfectly capable of line rate (at 100G+) IPv4 and IPv6 forwarding as well. When designing IPng Site Local, I created a number plan that assigns IPv4 from the 198.19.0.0/16 prefix, and IPv6 from the 2001:678:d78:500::/56 prefix. Within these, I allocate blocks for Continue reading

1 … 42 43 44 45 46 … 3,698