NANOG 78 Agenda
Agenda
Our 78th community-wide gathering was held February 10-12, 2020
Webcast
Watch all recorded talks on our YouTube + view the archived presentation decks.
View the WebcastView the Webcast on YouTube
Sunday, February 9, 2020
Topic/Presenter |
---|
Full AbstractBrought to you by Verizon Media with additional support from Tesuto, participants will have the opportunity to work with Panoptes (https://getpanoptes.io/), a global scale network telemetry ecosystem. Learn more at: https://www.nanog.org/meetings/nanog-78/hackathon/ |
|
Full AbstractKick off NANOG78 with EdgeConneX and its partners at #BeersWithPeers! Join us to network and enjoy complimentary cocktails, food and music at Barbarossa Lounge in San Francisco. **NANOG Badge required for entry** |
|
Monday, February 10, 2020
Topic/Presenter |
---|
|
Sponsors: |
Sponsors: |
NANOG 78 Conference Opening
L Sean Kennedy
Tom Daly - Fastly, Advisor
Vincent Celindro - Juniper Networks
Full AbstractJoin the NANOG Program Committee and Board of Directors Chairs, as well as our Conference Host Sponsor to kick off the 78th NANOG meeting. Speakers
|
RecordingsFull AbstractThe network is among the most critical components of any computing infrastructure. It is an enabler for modern distributed systems architecture with a trend toward ever-increasing functionality and offloads moving into the network. As such, it must continually be expanded and reconfigured to deploy compute and storage infrastructure. Most important, the network must deliver the highest levels of availability. Drawing from his experience with some of the largest networks at Google and driving vertical integration across large-scale compute, networking, and storage, Amin discusses the importance of network availability, the leading causes of failure, and the design principles key to delivering necessary levels of availability. Amin Vahdat |
Buffer Sizing and Video QoE Measurements at Netflix
Bruce Spang - Stanford University
Full AbstractIn this talk we will present some highlights from recent research results on how the sizing of router buffers affects Netflix video traffic. Our results were published at the recent Workshop on Buffer Sizing at Stanford (link to paper: http://buffer-workshop.stanford.edu/papers/paper12.pdf). In certain locations, Netflix streams video over TCP New Reno from racks of servers that are directly connected to large routers, which in turn directly peer with commercial ISPs. We varied the size of the router buffers during periods of persistent congestion, and logged metrics such as the number of rebuffering events, video quality, and video play delay. We observed buffers that are too small and too large, both of which worsen video QoE. Our main takeaways are: Speakers
|
|
Sponsors: |
The Next Network Professional: What's next?
David Temkin
Suzie Gleeson
Rami Rahim
Nick McKeown - Stanford University
Julia Stern
vijay gill - Google
Full AbstractAfter the success of the panel at NANOG 77, the number one piece of feedback I received was that the discussion was great, but people wanted more actionable advice - what could THEY do?. When I originally asked for an hour, I was afraid that we would run out of things to talk about - what happened in reality, was that we were only scratching the surface of what's important to the operator community. Part 2 of the panel focuses on what actionable things we can do - as leaders, as engineers, as partners, to include more people in our industry. We need to discuss what changes can we as people make, and what changes can we push our companies to make, to bring new, underrepresented faces in and help them flourish. Moderator: Dave Temkin, VP of Network and Systems at Netflix Speakers
|
SONiC - Enable Fast Evolution of Cloud Networking
Rita Hui - Microsoft
Full AbstractSONiC (Software for Open Networking in the Cloud) has been evolving fast. Built on top of SAI (Switch Abstraction Interface), SONiC is truly platform agnostic, enables its user to take full advantage of hardware innovations and keep the investment in the management system intact. SONiC’s unique containerized architecture plus Redis for state transition brings excellent extensibility to its users to customize for their scenarios. In this talk, we will present a full picture of SONiC to the audience – how it originated from challenges in hyper scale cloud networking, design concerns behind it, how the eco system evolved in the last two years, how Microsoft operates it and the roadmap in the near future. Speakers
|
|
Modern Cable Networks
Andy Smith - Cisco
Full AbstractThis talk is an overview of the strategy and application in the architecture of modern cable networks. While some of this content is specific to cable operators, the larger themes of packet and optical integration, automation, distribution of IP and Ethernet are applicable to service providers of any type or scope. Subjects include a discussion of the definition of network architecture, network discipline, technical advancements in silicon and optics and the evolution of cable metro and access networks. Speakers
|
DNS: The Protocols, The Myths, The Legends
Paul Ebersman - Neustar
Full AbstractThe DNS has been around for a long time. Over the last 35 years, But like all technology, the DNS must evolve. More widespread use This talk will cover some of the persistant bad information that Speakers
|
DNS Privacy in Practice - Measuring Deployment of DoT, DoH, and TFO
Casey Deccio - Brigham Young University
Full AbstractAn increased demand for privacy in Internet communications has resulted in privacy-centric enhancements to the Domain Name System (DNS), including the use of Transport Layer Security (TLS) and Hypertext Transfer Protocol Secure (HTTPS) for DNS queries. In this paper, we seek to answer questions about their deployment, including their prevalence and their characteristics. Our work includes an analysis of DNS-over-TLS (DoT) and DNS-over-HTTPS (DoH) availability at open resolvers and authoritative DNS servers. We find that DoT and DoH services exist on just a fraction of open resolvers, but among them are the major vendors of public DNS services. We also analyze the state of TCP Fast Open (TFO), which is considered key to reducing the latency associated with TCP-based DNS queries, required by DoT and DoH. The uptake of TFO is extremely low, both on the server side and the client side, and it must be improved to avoid performance degradation with continued adoption of DNS Privacy enhancements. Speakers
|
Tutorial: ARIN RPKI ROAthon
Mark Kosters - ARIN
Full AbstractIn this session, participants will learn about RPKI and the services offered by ARIN that support RPKI deployments. The workshop on Wednesday will offer hands-on lab exercises to create Route Origin Authorization objects (ROAs) within ARIN’s Operational Test & Evaluation Environment (OT&E). Speakers
|
Full AbstractThe forum provides time for attendees to meet and network with others in the peering community present at NANOG. Learn more at: https://nanog.org/meetings/nanog-78/peering-forum/ |
Full AbstractNANOG Social Event Location: SPIN Sponsors: Fastly & Windstream Wholesale **NANOG Badge required for entry** |
Tuesday, February 11, 2020
Topic/Presenter |
---|
|
|
Sponsors: |
|
RecordingsFull AbstractIn the last several decades, we have seen massive changes to networking and networking technology. From the hardware-dependent, scale up networks of then, to the software defined networks of now, cloud companies, service providers, and enterprises across the world have been on an exciting networking journey. Bikash Koley Prior to Google, he was the Executive Vice President and Chief Technology Officer of Juniper Networks. In this role, Koley charted Juniper’s technology strategy and led the execution of the company’s critical technology innovations. Specifically, he was responsible for Juniper’s telco cloud and virtualization, multicloud enterprise datacenter, and software-defined enterprise networking products and technologies. Prior to Juniper, Koley spent close to ten years at Google, where he was a Distinguished Engineer and the Head of Network Architecture, Engineering, and Planning. Prior to Google, he was the CTO of Qstreams Networks, a company he co-founded. Koley also spent several years at Ciena Corporation in various technical roles developing DWDM and Ethernet technologies. Koley is an industry-leading expert on network function virtualization, intent driven networking (IDN), multicloud networking, warehouse-scale computing, and hyperscale network infrastructures, and received a BTech from IIT, India; and MS and PhD degrees from the University of Maryland at College Park, all in Electrical Engineering. |
Recordings |
Connecting Indigenous Communities: Examples and Lessons Learned
Dr. Hosein Badran - Internet Society
Full AbstractIndigenous Communities across Canada and the US are among the most underserved in terms of Internet access. It is well established that reliable Internet service is a key enabler of economic development, entrepreneurship and SMEs, as well as much needed health and education services. Speakers
|
|
Sponsors: |
Lightning Talk: Internet Number Resource Fraud at ARIN
John Curran - ARIN
Full AbstractARIN's CEO John Curran will briefly discuss what qualifies as Internet number resource fraud at ARIN, how to report cases of suspected fraud, and how ARIN handles the investigation of these reports and follow-on activities. Speakers
|
Lightning Talk: RPKI Trust Anchor Usage and Cache Consistency
John Kristoff - NETSCOUT / Dataplane.org
Full AbstractRPKI ROAs are created and published into a handful of top-level trust anchors. Relying Party software periodically retrieves ROAs from the RPKI, validates them, and makes them available in a local cache for routers. ROA measurement studies and monitors have helped us to understand the data being put into the RPKI. Recent studies have also tried to measure the extent to which ROV is actively being deployed to However, little is known about the population of RPKI cache servers including synchronization patterns to the trust anchors and whether they have a reasonably consistent and complete set of valid ROAs. We aim to help fill this knowledge gap through our research. We are analyzing trust anchor access logs, measuring cache server consistency, and conducting route announcement experiments to better understand how the cache server infrastructure behaves in the real world. This talk will summarize our current progress to date, highlighting insights and challenges, as well as future directions. Most importantly, we are seeking network operator feedback and insight to help inform and improve our research. Speakers
|
Lightning Talk: Fighting BGP Route Leaks with PeeringDB’s new “never via route servers” flag
Theo Voss - ANEXIA Internetdienstleistungs GmbH
RecordingsFilesFull AbstractRoute servers are a convenience service that exists to lower the barrier to participate at an IXP. In the past, these route servers also distributed leaked routes from peers not participating and aggravated severe outages of the internet. Furthermore, the quality of BGP filters varies along IXPs. A few large operators implemented countermeasures like Peerlock but most other operators don't. With version 2.18.0, PeeringDB introduced a feature called “Never via route servers” for networks to indicate whether their routes should be reachable via route servers or not. This makes it possible to generate filters for all route server peerings and drop announcements containing AS numbers with “Never via route servers” flag in the AS path. Next to bogon filters, RPKI and IRR filters, this is another milestone in terms of automated routing security based on a central, authorized and well-maintained database. This talks explains how this can be easily used to generate filters by showing example API calls and router configuration. Speakers
|
Comparing the Network Performance of AWS, Azure, GCP, IBM Cloud and Alibaba Cloud
Angelique Medina - ThousandEyes
RecordingsFilesFull AbstractThis session will detail network performance and connectivity architecture variations between public cloud providers AWS, Microsoft Azure, Google Cloud, Alibaba Cloud, and IBM Cloud. Data was collected from global vantage points to cloud regions, within cloud backbones (inter-AZ and inter-region) and between different clouds. The presentation will highlight network performance differences and underlying causes, why Asia has the most variation in performance across the 5 cloud providers, how connectivity to/from China affects performance, and more. Speakers
|
Cloudflare - Anatomy of a Route Leak
Martin Levy - Cloudflare
Full AbstractOn June 24th, 2019 a route leak occurred like no other route leak beforehand. A combination of route optimizers, faulty IP filtering, and a series of phone calls to an unwitting NOC caused a large amount of traffic to be dropped for no good reason. This talk will show what happened and provide a good insight into how it could have been avoided. Speakers
|
Tutorial: IPv6 Intro
Alex Latzko - DEFT.COM
Full AbstractThis is brief introduction to IPv6 for those who need a refresher or are new to IPv6. Due to time limitations, we cannot cover the full breadth of supported features, but hopefully this is enough information to get the attendees started on their IPv6 journey! The topics we will cover are addressing concepts, differences between IPv4 and IPv6, some operational examples and things to look out for. Speakers
|
Full AbstractIt’s been 30 years since the famous meeting in a cafeteria where Kirk Lougheed and Yakov Rekhter used two napkins to sketch out the main feature of the BGP protocol. BGP was devised as an improved routing protocol able to fulfill the needs of an Internet that was about to take off. Yet as the architects themselves have admitted, security wasn't even on the table back then. And despite several security-driven protocol enhancements and BCPs over the past 30 years, thousands of companies and millions of end users are still affected by route leaks and hijack attempts that cause service disruptions and loss of revenue. In this talk, Catchpoint BGP expert Luca Sani will focus on route leaks and hijacks – he will explain what the biggest security risks are, how and why they came to be, and how they've affected end-user experiences around the world over the past year. Using one of the most famous leaks of 2019 as a case study, Luca will cover what went wrong and how it could have been prevented, hoping that such an analysis could raise awareness in the NANOG community on how important is to secure your network policies regardless of the size of your organization. |
Visualizing Major Routing Incidents in 3D
Doug Madory - Kentik
Full AbstractWhen we refer to a major routing leak, we often describe it simply with a single number: the number of unique prefixes mistakenly announced. However, this one-dimensional view of a complex incident obscures the fact that not every leaked route is in circulation for the same amount of time or propagated by the same number of ASes. This talk will describe a new approach for analyzing routing leaks using an interactive 3-dimensional visualization that attempts to capture these nuances of an incident. This talk will review major routing incidents from recent years and illustrate what this new approach to analyzing routing incidents reveals. Speakers
|
|
Intent Based Networking - the technology
Jeff Tantsura - Nvidia
Full AbstractWhat is the Intent Based Networking (IBN)? Intent defines the “what” not the “how”. In order to enforce that intent expectations are met, the IBNS has to be the single source of truth (regarding the intended state of both your infrastructure and your business rules) that one can programmatically reason about in the presence of change. Speakers
|
Full AbstractBGP policy misconfigurations have caused a number of headline grabbing outages lately. While there are some common sense improvements operators can make to help mitigate some of the causes, that is not enough. There is a need for comprehensive validation of routing policy at the peering edge. This is where Batfish comes into the picture. Batfish is an open-source network validation tool that builds models of routing and forwarding behavior of the network from the device configurations. Batfish enables operators to understand the impact of any configuration change before deploying it to the network. This talk will cover how Batfish works and demonstrate how an operator can validate a proposed change to BGP routing policy. |
Methods to Secure Routing
Rekha Rawat - Cisco
Mark Kosters - ARIN
John Kristoff - NETSCOUT / Dataplane.org
Nimrod Levy - AT&T
Doug Madory - Kentik
recorded
Speakers
|
Wednesday, February 12, 2020
Topic/Presenter |
---|
|
|
Sponsors: |
recorded
Speakers
|
the complexity of hyper speed transceivers - lets make it
Thomas Weible - Flexoptix
Full AbstractThomas will describe in detail the structures inside optical transceivers. A Transmitter / Receiver Optical Sub Assembly (TOSA / ROSA) is no longer just a diode in a housing handling the light path to and fro to the fiber. How did the optical industry players around the globe make it possible to squeeze everything into the tiny form factors we see today? It is all about precision - a microscope with a calm and competent hand is no longer sufficient, now it is about; nano tolerances, testing, complex transceiver firmware and a shed load of money. This is the high precision optical mechanical engineering revolution which fuels the hyper growth of data centers and optical networking worldwide… If you face design issues with your current optical network design Thomas will give insights into the latest 40G to 400G transceiver developments (e.g. long distance 80km) which you can expect to see in the upcoming months. Hopefully this might save you some headaches. As a small „one more thing" Thomas will dive into the basics of how FEC compensates for errors caused by PAM4 modulation. presenter: Thomas Weible - Co-Founder and CTO of Flexoptix GmbH. He formerly lead the groundbreaking software development within the company. Thomas has moved more and more towards the field of transceiver technology and his so called „support with no levels and no bullshit“. Enthusiastic in everything he does, he gives realistic and practical answers to get transceivers working and operational. As speaker at several conferences around the globe he is able to target the needs of network engineers. Speakers
|
ARIN RPKI ROA Hands On Session
Jon Worley - ARIN
Speakers
|
Full AbstractNetwork engineers understand that ASICs are the magical heart of what we do, but few of us understand how they work. Without violating any NDAs, this talk will shine some light on what ASICs are, how they operate and what their strengths and limitations are. We will briefly discuss single chip and multi-chip (chassis) systems and how they differ as well as the role buffers and table sizes play in the land of ASICs. |
IP/Optical awareness and correlation for traffic optimization
Filipe Correia - Ribbon
Full AbstractWith the introduction of 5G, IoT, Cloud infrastructure, and increased internet traffic demand, the network is going through a profound shift. Emerging applications and services will bring explosive growth in traffic volume in the near term, followed by traffic demands that become increasingly dynamic and elastic and require higher network resiliency along with continuous SLA monitoring and optimization across all layers in a transport infrastructure. The distinction between core, metro, and local area and treatment of traditionally siloed IP/Optical network layers will disappear over time. Today’s static networks, which require manual intervention to adapt to changes in services, bandwidth, and protection, will need to evolve to become a single automated network fabric built from awareness and correlation of network resources across all transport layers for more optimal placement, dynamic optimization, and increased reliability of ever demanding service and application traffic. Speakers
|
|
Public Speaking Forum
Christina Chu - NTT
Full AbstractThe Public Speaking Forum at NANOG 78 provides a positive and supportive environment for anyone interested in sharpening their public speaking + presentation skills, or sharing insights to help others sharpen theirs. It's also the perfect opportunity to dry run your next talk. Six 4-minute individual presentations will be given in the first hour of the forum, followed by table topics. The final 30 minutes are reserved for conversation + networking with other NANOG 78 attendees. Box lunches will be provided for the first 50 attendees. All levels welcome. Speaking slots are limited + first-come, first-served. Sign up at https://nanog.org/meetings/nanog-78/public-speaking-forum/ Speakers
|
Lightning Talk: Bridging the Gap between Industry and Collegiate Education in Networking
Tyler J Peatman
John Phan
RecordingsFilesFull AbstractThree Purdue University Cybersecurity/Network Engineering Undergraduate Students share their in-class Networking experiences as well as their industry experiences. We hope to convey a message that will bridge the gap between industry and the classroom. It is imperative to note that times are changing and that the computing industry is moving at a much quicker pace that computing education. With this being said, this presentation should encourage businesses to interact with collegiate students in computing disciplines to better prepare the future workforce. Presented by Tyler Peatman, John Phan and Ryan Tom Speakers
|
Lightning Talk: Monitoring your B and G and P
Chris Morrow - Google
Full AbstractShort talk on BGP/BMP monitoring: Speakers
|
Full AbstractAt Deutsche Telekom we recently created RQspec as tooling for making full combined use of old style IRR/RPSL and new RPKI/ROA information to evaluate routing policy databases. Speakers
|
Netdevops Survey : State of network operations through automation
Damien Garros - Network to Code
Full AbstractNetwork Automation has been a hot topic in the network industry for few years and yet we have very little data about the state of “network operations through automation" right now. In October 2019, we collected ~300 responses from all types of network that gave us some unique insight regarding what network engineer are doing when to comes to network automation, what tools they are using and how they managed their journey to network automation. This was the second edition of the survey so this time we also have data to understand the evolution overtime. This presentation will first present the results of the 2019 edition and share the most interesting data. In a second part, I’ll present the Netdevops Survey project, how to get involved, provide feedback and access the results Speakers
|
Full AbstractFrom physical to virtual to the cloud (and now multi-cloud), networks are getting more diverse. For network professionals, managing across all of the diversity is a growing challenge. On top of that, SDN and SD-WAN are added architecture elements making the network stack even more difficult to holistically understand. Add virtualization, network overlays, and container networks to the mix, and you’ve got increasingly dynamic networks — i.e. networks that come and go — which make network visibility and management harder than ever to achieve. If you couple this with vendor-specific tools, the chore of managing a portfolio of network management software has become increasingly challenging. What’s the answer to managing today’s networks? Automation. However, automation does not necessarily mean simplicity. Most network teams have countless automation tools, and many are adding more. As teams transition from custom tools to network configuration and change management (NCCM), fast forward to DevOps tools, and throw in the future promise of intent-based networking, how is a team to manage this level of complexity? In his talk, former Gartner analyst Jonah Kowall, CTO of Kentik, will discuss the critical role that network teams play in automation. He will explain why the current promise (or “hype”) of network automation is centered around closed-loop automation (a combination of telemetry, analytics, and orchestration) to drive a future network which incorporates AIOps platforms to integrate different technologies in a more repeatable way to operate a network and facilitating self-healing and scaling networks. At the same time, he'll cover where we truly are: between the promise of automation and the baseline of ad-hoc coding of scripts for specific workflows. Additionally, Kowall will offer advice on what’s needed in order for the industry to make the leap from partial to full network automation. The audience will walk away with a better understanding of the current automation strategies that can be applied within their own organizations, in addition to learning what’s ahead for network automation. |
An Open Platform to Teach How the Internet Practically Works
Thomas Holterbach - ETH Zurich
Full AbstractIn this talk, we would like to present a platform that we have been using in the last four years at ETH Zurich to teach our students how the Internet practically works. Our platform faithfully emulates the real Internet infrastructure and allows our students to operate their very own Internet infrastructure composed of hundreds of routers and dozens of Autonomous Systems (ASes). Their goal? Enabling Internet-wide connectivity. Speakers
|
Sponsors: |
RecordingsFilesFull AbstractEfficiently planning the wide area network (WAN) provides strategic value to the operator. The python3 Network Traffic Modeler (pyNTM) is an open source network simulation engine. Users with basic to intermediate python experience and a reasonable traffic matrix now have the ability to run simulations that allow them to gain understanding of their network and how to more efficiently grow it without overbuilding (stranding capital) or underbuilding (increased risk). This talk will cover basic network modeling concepts, the strategic value of modeling the WAN, and how pyNTM facilitates effective planning and understanding of the WAN. It will also cover pyNTM's place in the modeling ecosystem and what types of organizations pyNTM will create value for. Agenda: The demo will include Speakers
|
Full AbstractThe SRv6 network programming model was first presented at IETF in March 2017. During the last two years, the ecosystem around the technology has made tremendous and successful implementation, deployment and standardization efforts. In this talk, we will present the SRv6 ecosystem, standardization progress and the available implementations, both open source and commercial. Then we will delve into the SRv6 deployed use-cases from (Softbank, Iliad) and planned deployments. |
NANOG 78 Conference Closing
Vincent Celindro - Juniper Networks
FilesSpeakers
|
Network Lounge: Netrality
Service Sponsors: ARIN, Cloudflare, ISC, ServerCentral