In the second half of his Networks, Buffers, and Drops webinar JR Rivers focused on end systems: what tools could you use to measure end-to-end TCP throughput, or monitor performance of an individual socket or the whole TCP stack?

REST API is the way of the world and all network devices should support it, right? Well, Ken Duda (Arista) disagreed with this idea during his Networking Field Day presentation, but unfortunately there wasn’t enough time to go into the details that would totally derail the presentation anyway.

Fixing that omission: should we have REST API on network devices or not?

My BGP in EVPN-Based Data Center Fabrics blog post generated numerous comments from engineers disagreeing with my views on using IBGP-over-EBGP.

As usual, there were three kinds of comments:

The idea of generating random IPv6 addresses (so you cannot be tracked across multiple networks based on your MAC address) that stay stable within each subnet (so you don’t pollute everyone’s ND cache every time you open your iPad) is pretty old: RFC 7217 was published almost exactly four years ago.

Linux was quick to pick it up, OpenBSD got RFC 7127 support a few weeks ago. However, there’s an Easter egg in the OpenBSD patches that implement it: SLAAC on OpenBSD now works with any prefix length (not just /64).

Years ago Petr Lapukhov decided that it’s a waste of time to try to make OSPF or IS-IS work in large-scale data center leaf-and-spine fabrics and figured out how to use BGP as a better IGP.

In the meantime, old-time routing gurus started designing routing protocols targeting a specific environment: highly meshed leaf-and-spine fabrics. First in the list: Routing in Fat Trees (RIFT).

One of my readers found a Culumus Networks article that explains why you can’t have more than a few hundred VXLAN-based VLAN segments on every port of 48-port Trident-2 data center switch. That article has unfortunately disappeared in the meantime, and even the Wayback Machine doesn’t have a copy.

Andy sent me this question:

I'm currently playing around with BGP & VXLANs and wondering: is there anything preventing from building a virtual IXP with VXLAN? This would be then a large layer 2 network - but why have nobody build this to now, or why do internet exchanges do not provide this?

There was at least one IXP that was running on top of VXLAN. I wanted to do a podcast about it with people who helped them build it in early 2015 but one of them got a gag order.

The pace of live webinar sessions will slow down a bit in April 2018 due to the onslaught of European spring holiday season. Nonetheless, you’ll be able to enjoy:

• The second part of EVPN Technical Deep Dive series with Dinesh Dutt on April 5th;
• The planning and design part of NSX, ACI or EVPN webinar with Mitja Robas on April 24th;

On April 19th we’ll have the first DIGS event in 2018, starting with introduction to SDDC and VMware NSX in the morning and NSX workshop in the afternoon.

Sitting in a taxi driving to CLEUR 2018 in Barcelona we couldn’t resist but complain about the stuff we’re seeing in real-life networks, resulting in someone exclaiming something along the lines of “I can’t understand how someone could do so many stupid things”

Welcome to the wonderful world of Dunning-Kruger Effect.

The networking engineers attending the Building Network Automation Solutions online course created numerous amazing automation solutions, most of them already deployed in production networks.

I described some of them in my Troopers 2018 Real-Life Automation Wins talk. The presentation is online and the video has been published on YouTube a few days ago. I hope you’ll find it as inspirational as the Troopers attendees did.

Did you create an awesome automation solution? I’d like to hear about it!

This blog post was initially sent to the subscribers of my SDN and Network Automation mailing list. Subscribe here.

Alex was trying to figure out how to use Catalyst 3850 switches and sent me this question:

Is MLAG an alternative to use rather than physically creating a switch stack?

Let’s start with some terminology.

Link Aggregation Group (LAG) is the ability to bond multiple Ethernet links into a single virtual link. LAG (as defined in 802.1ax standard) can be used between a pair of adjacent nodes. While that’s good enough if you need more bandwidth it doesn’t help if you want to increase redundancy of your solution by connecting your edge device to two switches while using all uplinks and avoiding the shortcomings of STP. Sounds a bit like trying to keep the cake while eating it.

When VMware launched the first version of NSX for vSphere more than four years ago, the NSBU team reached out to me and asked me to create a sponsored webinar describing NSX fundamentals, its architecture, and high-level deployment guidelines.

In the meantime we discussed updating the materials, but nothing ever happened. Time to fix that, this time from a vendor-neutral perspective. We’ll start with a day-long event on April 19^th 2018 in Zurich, Switzerland.

I asked David Gee to review my streaming telemetry blog posts to make sure I didn’t make too many blunders, and he sent me a nice summary of his view on the topic in return.

The only thing I could do after reading it was to ask him for permission to do a copy-paste. Here it is:

One of the biggest challenges of network automation is getting usable information from network devices… or as asked by a student in my Building Network Automation Solutions online course in the course Slack team:

How do I get specific information from a specific command from a device without an Ansible Network Module? Is Python the only suggested approach?

I described how hard it is to get structured information from network devices in great details in this section of the Ansible for Networking Engineers webinar and online course. Here are a few more thoughts on the topic:

Someone pointed me to this article by dr. Paul Vixie (of the DNS fame). The best part (as I’m not a security person):

The TCO of new technology products and services, including security-related products and services, should be fudge-factored by at least 3X to account for the cost of reduced understanding. That extra 2X is a source of new spending: on training, on auditing, on staff growth and retention, on in-house integration.

In case you didn’t get it: figure out how much you think the magic unicorn-based software-defined solution will cost, then multiply it by three. Of course nobody wants to admit that.