datacenter networking

John Ousterhout's Netdev 0x16 keynote arguing that nearly every aspect of TCP is wrong for the datacenter, from connections and streams to reliable delivery and congestion control. He presents the case for a replacement protocol.

Google's Amin Vahdat gives the ONS 2015 keynote on Google's datacenter networking, including its custom hardware, software-defined networking and very high bisection bandwidth.

Van Jacobson's 2006 Google Tech Talk argues that networking should be rethought around named data rather than host-to-host conversations. He contrasts the circuit and datagram views of the internet and sketches content-centric networking as a replacement for the location-centric TCP/IP model.

Van Jacobson's 2006 Google Tech Talk argues for rethinking networking around named data rather than host-to-host connections. He gives the history and the unsolved problems of the current internet architecture.

A two-hour NANOG tutorial by Richard Steenbergen aimed at network engineers, covering optical networking technologies. This is the updated 2017 edition of a recurring talk.

Scott Manley explains the thermal balance of spacecraft and how a satellite could be configured to radiate away the heat from datacenter-scale computing in space. His rough numbers suggest it is feasible.

James Hamilton of AWS gives a Velocity 2010 keynote on datacenter infrastructure innovation, focused on power and cooling costs and a breakdown of where datacenter spending goes.

Bungie's David Aldridge explains the networking programming behind Halo: Reach in a 2011 GDC session. It covers how gameplay stays in sync online, including forecasting and state reconciliation.

Urs Hoelzle, Google's SVP of infrastructure, gives the 2012 Open Networking Summit keynote on how Google adopted OpenFlow and software-defined networking in its own networks. The slides are linked in the description.

Van Jacobson's 2006 Google Tech Talk on rethinking networking around named content instead of host-to-host conversations. He recounts packet-networking history and lays out the content-centric networking idea that later became NDN.

Eric Brandwine's AWS re:Invent 2015 session on how Amazon VPC networking works. It covers the problems the design solved, why traditional approaches do not scale, and the mechanisms used for customer isolation and VPC endpoints.

James Hamilton's Tuesday Night Live keynote at AWS re:Invent 2016, touring AWS infrastructure at scale including networking, power and custom silicon. It also features NASA JPL's Tom Soderstrom.

James Hamilton, AWS VP and distinguished engineer, gives an insider view of the innovations behind the AWS cloud at re:Invent 2014. He covers custom hardware, networking, regions and availability zones, and infrastructure at scale.

A 2018 GDC talk by NetherRealm's Michael Stallone on the engine changes, optimizations and tooling needed to roll back and re-simulate up to 8 frames in 16 ms in Mortal Kombat and Injustice 2. It is a deep dive into deterministic rollback netcode.

Benn Jordan investigates infrasound and low-frequency noise from data centers, including xAI's Colossus and a crypto mine, and reports his own experiments on the effects. The video covers the health and community impacts on nearby residents.

A two-minute 2008 clip from Sun's Fishworks team in which Brendan Gregg uses DTrace to show disk latency spiking while he shouts at a rack of drives. Bryan Cantrill narrates the demonstration.

A short video of five people carrying a live, running web server across the city by public transport to a new datacenter. It stayed powered and connected to the internet the whole way, so there was no downtime.

Bryan Cantrill of Oxide Computer Company gives a Stanford EE380 talk on rethinking the server. He argues that hyperscaler-style rack-scale machines with open firmware, RISC-V and Rust should be available to everyone, not just the giants.

Jared Mauch tells how he got fed up with poor internet options in rural Michigan and built his own fiber network to his town. The talk was given at NLNOG Live in September 2020.

Mark Handley of UCL explains, with simulation visualizations, how early Starlink satellites without inter-satellite links could still give low-latency wide-area communication by hopping via ground relays. He shows it can beat fibre on many long paths.

Pat Wyatt, a veteran game programmer who worked at Blizzard and ArenaNet, speaks at HandmadeCon 2015 in a conversation with Casey Muratori. He discusses network code and architecture in games he worked on, including mistakes on Diablo and the success of Guild Wars.

Netflix engineer Drew Gallatin's EuroBSDcon 2022 talk quantifies the older FreeBSD optimizations that let Netflix serve web traffic at 800Gb/s, including TSO, LRO, VM page cache zones, asynchronous sendfile and unmapped mbufs.

A NANOG 64 panel with speakers from Comcast, APNIC, T-Mobile and Facebook on deploying IPv6-only networks, including performance data showing IPv6 outperforming NAT-ed IPv4.

Andreas Spiess explains the basic principles of LoRa and LoRaWAN, their pros and cons, link budget and bandwidth. He shows his own gateway and client sending a message to The Things Network.

Jon Gjengset's roughly four-hour live stream implementing part of a BitTorrent client in Rust, following CodeCrafters' test-guided challenge. He builds from scratch and shows idiomatic error handling along the way.

Branch Education explains how Starlink works, from the ground dish and satellites to phased-array beam steering and data routing. Animations cover how a beam stays connected to a satellite moving at 27,000 km/h.

Wells Fargo engineers walk through the bank's four-year IPv6 journey at the North American IPv6 Summit 2014. They cover motivation, approach, achievements and observations.

Netflix engineer Drew Gallatin explains how Netflix serves TLS-encrypted video at 400Gb/s from a single FreeBSD server. He covers NUMA, kernel TLS, hardware TLS offload and the memory bandwidth and PCIe bottlenecks they hit.

Cloth Map documents Havana's SNET, a community-built private network that Cubans created because internet prices were high. It is used for gaming and sharing.

Bryan Cantrill's GOTO Chicago 2017 talk on debugging production systems under pressure, centered on the incident where an operator's mistyped command rebooted an entire Joyent datacenter. It covers human factors, automation and architecture.

Martin Kleppmann's GOTO Berlin 2016 talk on how collaborative editors, offline apps, and multi-datacenter databases handle concurrent updates. It covers conflict resolution approaches including operational transformation and CRDTs, with a worked look at text CRDTs.

Bryan Hunter's GigCityElixir 2023 talk on HCA Healthcare's Waterpark system, built on Elixir/Erlang. It describes how the architecture is distributed and fault-tolerant, with a multi-datacenter cluster and runtime code swapping.

A demonstration of the 1982 Xerox Star user interface, including its desktop, icons, documents, networking and printing. The uploader argues that Apple's later Lisa and Mac interfaces evolved from these ideas rather than copying them.

Kirk McKusick gives a narrative history of the Berkeley Software Distributions at MeetBSD California. He covers how BSD originated and evolved, including the origin of the networking releases.

Second half of a demonstration of the 1982 Xerox Star user interface. It shows how the Star differed from the later Lisa and Mac interfaces, with strengths in networking and printing and little direct manipulation.

A one-minute clip from the 1995 Steve Jobs interview in which he describes what he saw at Xerox PARC and what he missed. He says he grasped the graphical interface but not the networking and object-oriented parts.