Category: networking fundamentals

A subscriber sent me this intriguing question:

Is it not theoretically possible for Ethernet frames to be 64k long if ASIC vendors simply bothered or decided to design/make chipsets that supported it? How did we end up in the 1.5k neighborhood? In whose best interest did this happen?

Remember that Ethernet started as a shared-cable 10 Mbps technology. Transmitting a 64k frame on that technology would take approximately 50 msec (or as long as getting from East Coast to West Coast). Also, Ethernet had no tight media access control like Token Ring, so it would be possible for a single host to transmit multiple frames without anyone else getting airtime, resulting in unacceptable delays.

Grouping the features needed in a networking stack in a bunch of layered modules is a great idea. Unfortunately, you could place several essential features like error recovery, retransmission, and flow control in several different layers, from the data link layer dealing with individual network segments, to the transport layer dealing with reliable end-to-end transmissions.

Where should we put those modules? As always, the correct answer is it depends, in this particular case, on transmission reliability, latency, and bandwidth cost. You’ll find more details in the Retransmissions and Flow Control part of How Networks Really Work webinar.

Two weeks ago I replied to a battle-scar reaction to 7-layer OSI model, this time I’ll address a much more nuanced view from Russ White. Please read his article first (as always, it’s well worth reading) and when you come back we’ll focus on this claim:

The OSI Model does not accurately describe networks.

Like with any tool in your toolbox, you can view the 7-layer OSI model in a number of ways. In the case of OSI model, it can be used:

After identifying some of the challenges every network solution must address (part 1, part 2, part 3) we tried to tackle an interesting question: “how do you implement this whole spaghetti mess in a somewhat-reliable and structured way?”

The Roman Empire had an answer more than 2000 years ago: divide-and-conquer (aka “eating the elephant one bite at a time”). These days, we call it layering and abstractions.

In the Need for Network Layers video, I listed all the challenges we have to address and then described how you could group them in meaningful modules (called networking layers).

In the introductory videos of How Networks Really Work webinar I described the mandatory elements of any networking solution and additional challenges you have to solve when you can’t pull a cable between the adjacent nodes.

It’s time for the next bit of complexity: what if we have more than two nodes connected to the same network segment? Welcome to the world of multi-access networks and data link control.

After discussing the challenges one encounters even in the simplest networking scenario connecting two computers with a cable, we took a short diversion into an exciting complication: what if the two computers are far apart and we can’t pull a cable between them?

Trying to answer that question, we entered the wondrous world of transmission technologies. It’s a topic one can spend a whole life exploring and mastering, so we were not able to do more than cover the fundamentals of modulations and multiplexing technologies.

Whenever discussing a complex topic, it’s worth adhering to two principles: (A) identify the challenges you’re trying to solve, and (B) start as simple as you can and add complexity later.

We did precisely that in the Introducing Networking Challenges part of How Networks Really Work webinar. We started with the simplest possible case of two computers connected with a cable… and even there identified a plethora of challenges that had to be solved more than half a century ago (and still have to be solved today no matter what magic software-defined technology someone pulls out of their wizard hat).

In mid-June I started another pet project - a series of webinars focused on networking fundamentals. In the first live session on June 18th we focused on identifying the challenges one has to solve when building an end-to-end networking solution, and the role of layered approach to networking.

Not surprisingly, we quickly went down the rabbit holes of computer networking history, including SCSI cables, serial connections and modems… but that’s where it all started, and some of the concepts developed at that time are still used today… oftentimes heavily morphed by recursive application of RFC 1925 Rule 11.

I was speaking with a participant of an SDN event in Zurich after the presentations, and he made an interesting comment: whenever he experienced serious troubleshooting problems in his career, it was due to lack of understanding of networking fundamentals.

Let me give you a few examples: Do you know how ARP works? What is proxy ARP? How does TCP offload work and why is it useful? What is an Ethernet collision and when would you see one? Why do we need MLD in IPv6 neighbor discovery?

Every single network device (or a distributed system like QFabric) has to perform at least three distinct activities:

• Process the transit traffic (that’s why we buy them) in the data plane;
• Figure out what’s going on around it with the control plane protocols;
• Interact with its owner (or NMS) through the management plane.

Routers are used as a typical example in every text describing the three planes of operation, so let’s stick to this time-honored tradition:

During the last weeks I tried hard to sort out my thoughts on routing and bridging; specifically, what’s the difference between them and why you should use routing and not bridging in any large-scale network (regardless of whether it happens to be cramped into a single building called Data Center).

My vague understanding of layer 2 (Data Link layer) of the OSI model was simple: it was supposed to provide frame transport between neighbors (a neighbor is someone who is on the same physical medium as you are); layer 3 (Network layer) was supposed to provide forwarding between distant end nodes. Somehow the bridges did not fit this nice picture.

As I was struggling with this ethereally geeky version of a much older angels-on-a-pin problem, Greg Ferro of EtherealMind.com (what a coincidence, isn’t it) shared a link to a GoogleTalk given by Radia Perlman, the author of the Spanning Tree Protocol and co-author of TRILL. And guess what – in her opening minutes she said “Bridges don’t make sense. If you do packet forwarding, you should do it on layer 3”. That’s so good to hear; I’m not crazy after all.

IS-IS has “forever” (at least since RFC 1195) supported multiple layer-3 protocol, but always with a nasty side-effect: if a link in your network did not support one of them, you could get hard-to-diagnose black holes. The problem is illustrated in the left-hand column of the following diagram. Due to a single IS-IS topology, the shortest path between A and B is the direct link and since IPv6 is not enabled on that link (click on the diagram to get an enlarged version where you'll be able to see the link colors), A and B cannot exchange IPv6 traffic even though there’s an alternate path between them.

You might think that the lack of a decent session layer in the TCP/IP protocol suite is the main culprit for our reliance on IP multihoming and related explosion of the IP routing tables. Unfortunately, we have an even bigger problem: the Berkeley Socket API, which is around 40 years old and used in almost all TCP/IP software implementations and clients (including high-level scripting languages like PERL or Python).

One of the biggest challenges facing the Internet core today is the explosion of the IP routing and forwarding tables, which is caused primarily by traffic engineering and multihoming requirements. Things were supposed to get better when IPv6 introduced strict hierarchical addressing (similar to the phone number addressing, where the first few digits always denote the country code).

Unfortunately, the hierarchical IPv6 addressing idea relied on incredible belief that the world will shape itself according to the wills of the IETF working group members. Not surprisingly, that didn’t happen and the hierarchical IPv6 addressing idea was quietly scrapped, giving us plenty more prefixes to play with when trying to pollute the global IPv6 routing tables.