Every now and then I find an IT professional claiming we should not be worried about split-brain scenarios because you have redundant links.

I might understand that sentiment coming from software developers, but I also encountered it when discussing stretched clusters or even SDN controllers deployed across multiple data centers.

Finally I found a great analogy you might find useful. A reader of my blog pointed me to the awesome Why Must Systems Be Operated blog post explaining the same problem from the storage perspective, so the next time you might want to use this one: “so you’re saying you don’t need backup because you have RAID disks”. If someone agrees with that, don’t walk away… RUN!

Got this question from one of ipSpace.net subscribers:

Do we really need those intelligent datacenter switches for underlay now that we have NSX in our datacenter? Now that we have taken a lot of the intelligence out of our underlying network, what must the underlying network really provide?

Reading the marketing white papers the answer would be IP connectivity… but keep in mind that building your infrastructure based on information from vendor white papers usually gives you the results your gullibility deserves.

January 2020 was one of the busiest months we ever had:

• Elisa Jasinska and Nick Buraglio started the year with the first part of Surviving in Internet Default-Free Zone webinar.
• Krzysztof Szarkowicz completed the EVPN-with-MPLS topic describing the intricacies of layer-2 and layer-3 integration, and extending the EVPN webinar to almost 11 hours (with three hours of it covering EVPN in SP/MPLS world).
• Matthias Luft continued the Cloud Security saga, covering logging, monitoring, testing and auditing.
• Preparing for our public cloud course, I described how you can automate AWS infrastructure deployments.
• Christopher Werny wrapped up the month with the second part of IPv6 Enterprise Security talk, this time focusing on layer-2 security.

You can get immediate access to all these webinars with Standard or Expert ipSpace.net subscription.

During a recent workshop I made a comment along the lines “be careful with feature X from vendor Y because it took vendor Z two years to fix all the bugs in a very similar feature”, and someone immediately asked “are you saying it doesn’t work?”

My answer: “I never said that, I just drew inferences from other people’s struggles.”

A Step Back

Networking operating systems are probably some of the most complex pieces of software out there. Distributed systems are hard. Real-time distributed systems are even harder. Real-time distributed systems running on top of eventually-consistent distributed databases are extra fun.

After introducing the fallacies of distributed computing in the How Networks Really Work webinar, I focused on the first one: the network is (not) reliable.

While that might be understood by most networking professionals (and ignored by many developers), here’s an interesting shocker: even TCP is not always reliable (see also: Joel Spolsky’s take on Leaky Abstractions).

A journey of a thousand miles begins with one step they say… but what should that first step be if you want to start a network automation journey (and have no idea how to do it)?

Anne Baretta sent me a detailed description of his journey, which (as is often the case) started with the standardized configuration templates.

Aldrin wrote a well-thought-out comment to my EVPN Dilemma blog post explaining why he thinks it makes sense to use Juniper’s IBGP (EVPN) over EBGP (underlay) design. The only problem I have is that I forcefully disagree with many of his assumptions.

He started with an in-depth explanation of why EBGP over directly-connected interfaces makes little sense:

If you’re an ipSpace.net subscriber, you might have noticed how busy the last month has been (more about that later). February won’t be much better:

• Later today we’ll have David Barroso talk about safely managing network automation secrets.
• On February 6th I’ll describe the tools you can use to automate Azure deployments, including simple CLI scripts, Ansible, Terraform, and Azure Resource Manager templates.
• We’re starting the Networking in Public Cloud Deployments online course on February 11th.
• David Peñaloza Seijas will talk about Cisco (Viptela) SD-WAN on February 13th;

Finally, I’ll run a day-long workshop in Zurich on March 10th describing containers and Docker.

Unless you’re working for a cloud-only startup, you’ll always have to connect applications running in a public cloud with existing systems or databases running in a more traditional environment, or connect your users to public cloud workloads.

Public cloud providers love stable and robust solutions, and they took the same approach when implementing their legacy connectivity solutions: you could use routed Ethernet connections or IPsec VPN, and run BGP across them, turning the problem into a well-understood routing problem.

In January 2020 Doug Heckaman documented his experience with VeloCloud SD-WAN. He tried to be positive, but for whatever reason this particular bit caught my interest:

Edge Gateways have a limited number of tunnels they can support […]

WTF? Wasn’t x86-based software packet forwarding supposed to bring infinite resources and nirvana? How badly written must your solution be to have a limited number of IPsec tunnels on a decent x86 CPU?