We are excited to announce the CHI-NOG 09 agenda. With the conference coming up next month we have a great agenda with spectacular speakers coming from all over the world to Chicago bringing our community together. This year’s program focuses mainly on network automation, but also has very unique topics on AI with network operations, SDN architectures, public cloud performance, network monitoring, open networking, latest updates in IGPs.
|Opening Remarks by Tom Kacprzynski
|Why Entangle? How ? How Many Hops? (Link-state Protocol Support for Multihop Entanglement) by Kireeti Kompella (Slides)
|Applying Machine Learning to Networks by Rahul Aggarwal (Slides)
|Introduction to Modern Time Series Database for Network Engineer by Damien Garros (Slides)
|Powering Your Automation: A Single Source of Truth by Tim Schreyack (Slides)
|Chemical Kinetics in Computer Networks by Rachel Traylor (Slides)
|Disaggregation and Business Value by Russ White (Slides)
|SDWAN Architecture Concepts by Ali Shaikh (Slides)
|Neglecting Automated Testing Or: How To Take Down Your Network In Three Easy Steps by Anthony Miloslavsky (Slides)
|Automating a Multi-Vendor Network by Teren Sapp (Slides)
|IGPs – The New Hotness by Steve Ulrich (Slides)
|How to get started with IETF (Panel) with Joe Clarke,
Nagendra Kumar Nainar, Russ White
|Comparing the network performance of AWS, Azure and GCP by Yefim Pipko (Slides)
|Using YANG Models and Telemetry for Closed-Loop Applications by Craig Hill (Slides)
|Getting to Know Model-Driven Management With the YANG Catalog by Joe Clarke (Slides)
|Seamless BFD (S-BFD) – Scalable Path Failure Detection Mechanism by Nagendra Kumar Nainar (Slides)
|Steelcase – Deployment of Open Networking & SDN in a Large Enterprise by Scott Schafer (Slides)
|It’s 2019 and You are Still Getting Hacked by Otha Mabry
|Reception / Social
Please see the detailed agenda below, but don’t forget to register soon as regular ticket sale end this month.
Link-State Protocol Support for Multihop Entanglement by Kireeti Kompella (Juniper)
This talk gives a very brief description of Quantum Entanglement, how it’s produced, and why it’s interesting. Then the operations of entanglement distillation and entanglement swap are discussed. These operations are used to create a long distance entanglement, given the ability to create several shorter range entanglements. Finally, the role of ISIS/OSPF in enabling and enhancing this process will be described.
Applying Machine Learning to Networks by Rahul Aggarwal (Augtera Networks)
Operational complexity in networks continues to increase particularly with growing adoption of SDN, multi-cloud and white box switching platforms. As a result there is increasing need for providing machine based operational intelligence in networks to help humans keep up with the complexity. However applying machine learning to networks has significant technology challenges. We will go over these challenges and the fundamental technology innovation that is required across multiple domains to overcome these challenges. We will also present the learnings and results from deploying Machine Learning based software in large production networks.
Disaggregation and Business Value by Russ White (Juniper)
Disaggregation began with the hyperscalers, and is quickly moving through the rest of the network engineering world. What is disaggregation, really? Does it always mean open source and white boxes, or is there some broader meaning that can be applied across many different products to build and add value to many different kinds of networks? How does disaggregation add business value above the current model of purchasing, racking, stacking, and configuring appliance based network devices? This session will delve into the world of disaggregation, helping you understand how and where disaggregation intersects with business agility and treating the network as a value add, rather than a bit transport commodity.
Introduction to Modern Time Series Database for Network Engineerby Damien Garros (Roblox)
Telemetry / Monitoring, also known as Observability, as been a hot topic in the software industry for few years. With so many moving pieces, the mindset around telemetry shifted from I know what I want to monitor to I don’t know yet what might be important to understand the next outage. It seems obvious but previously the tools available to monitor applications and system had some limitations that would limit how much we can explore the data after the fact but most importantly how much we could analyze, correlate or compare different datasets. This new requirement drove a new set of tools that changed the paradigme around monitoring. The goal of this talk is to help Network Engineer to get started with modern Time serie database (influxdb, prometheus) and associated visualization tools like Grafana.
Powering Your Automation: A Single Source of Truth by Tim Schreyack (Virtustream)
Without a single, dynamic, and authoritative source for data, your automation is fragile, prone to error and not likely not idempotent. Here’s a look at how you can change the way you store your data and shift everything (logical and physical) to one central database, which in turn can be used to make your automation go ’round.
Using YANG Models and Telemetry for Closed-Loop Applications by Craig Hill (Cisco)
This discussion will focus around automation and programmability solutions that leverage closed-loop techniques for applications (security and others) requiring near real-time reactions to changes and/or anomalies in the network. The solution covered in this talk will target a suite of open source programmability (YANG Development Kit – YDK) using YANG models and streaming telemetry solutions (Kafka, Pipeline) beyond legacy SNMP like functions. This session will also provide a demonstration of the solutions, protocols, and methods discussed.
Automating a Multi-Vendor Network by Teren Sapp (Arista)
Building a multi-vendor network has many advantages however one of the primary drawbacks is how does an organization manage a multi-vendor network in an effective manner? There are plenty of tools out there however Napalm and Nornir were created with network operators in mind to abstract the vendor command line interface or API and give operators a single point to issue configuration and gather statistics from network devices. In this talk we’ll look through the inner workings of Napalm and Nornir and show how these tools can simplify and automate tasks in a multi-vendor network.
SDWAN Architecture Concepts by Ali Shaikh (Cisco)
A discussion on separation of control and data planes principles in SDN with an emphasis on the need for greater cryptographic strength, scale and the ability to build arbitrary networks. This will include a conversation on how implementations for SDN for WAN networks need to employ a true separation of the control plane and data plane that allows for SDN controllers to be responsible for routing policies and fabric creation, while the edge endpoints become dataplane enforcement points. This will be expanded on by exploring the practical methods to implement strong security and cryptographic methods, PKI, AES-256 encryption to build SDWAN fabrics using the highest grade encryption and how to build large scale IPSec fabrics, an SDN control plane is required. Finally the session will explore practical considerations when building large BGP style networks over the WAN and maintaining VRF style segmentation without inheriting the complexities of classical point router implementations and leveraging SDN methods to build centralized routing frameworks.
IGPs – The New Hotness by Steve Ulrich (Arista)
There’s been a resurgence in activity in the IGP domain, mostly by folks looking to do interesting things in the data center and in deploying new service provider architectures. For some time, folks have been (ab)using BGP in the data center to address link-state flooding scale. This has triggered new work in the IETF and other organizations to enhance link-state routing protocols to address some of these concerns in DC environments and to improve the efficiency of new service provider architectures. This talk will briefly discuss some of the approaches being taken, the objectives of some of these approaches and some of the relative merits of these approaches.
Neglecting Automated Testing Or: How To Take Down Your Network In Three Easy Steps by Anthony Miloslavsky (Cumulus)
As the idea of treating your infrastructure as code (via CI/CD) becomes more and more popular in the networking space, the testing and validation best practices that are present in traditional software development pipelines are often an afterthought or a “nice to have”. After some background on why automation principles in networking both are and aren’t different from those in software development, we’ll touch on the various types of testing that are present in software development today (unit, integration, etc) and how they can be applied to automating your network. The discussion will focus on developing reusable post-change testing, and will end with incorporating network simulation into your pipelines to achieve true pre and post-change testing.
Comparing the Network Performance of AWS, Azure and GCP by Yefim Pipko (ThousandEyes)
Amazon Web Services, Microsoft Azure and Google Cloud all significantly invest in their backbone networks and global regional data center presence to support performance requirements. So how do these cloud networks actually behave and perform? This session will discuss the network performance and connectivity architecture findings gained from global measurements of Amazon Web Services, Microsoft Azure and Google Cloud, collected from global vantage points to cloud regions, within cloud backbones(inter-AZ and inter-region) and across clouds (multi-cloud). We’ll also share some interesting anomalies we observed in connectivity and performance stability, particularly in the Asia-Pacific region.
Getting to Know Model-Driven Management With the YANG Catalog by Joe Clarke (Cisco)
How many people think the CLI is an API? Sadly it’s been used like one for a long time. Screen-scraping has become an art form and a never-ending source of frustration. Fortunately, data model-driven management has been on the rise of late. These data models, typically created using the YANG data modeling language, enable more powerful configuration automation and richer operational insights through machine-consumable structures and streaming telemetry. But how can you get from CLI and SNMP management into data-driven management? What are the tools that allow an operator to find the right YANG module to use? The YANG Catalog is an open source answer to these questions by providing YANG module search capabilities, metadata exploration, and the ability to generate sample code (to name a few). This talk introduces the YANG Catalog to give operators and engineers another tool that will help them better automate and control their network.
Chemical Kinetics in Computer Networks by Rachel Traylor (Marquette University)
In an effort to bring academic research to the private sector, this talk will give a tutorial and discussion on recent research into a natural form of flow control that is based on chemical kinetics. We will discuss the notion of an artificial packet chemistry and its use in scheduling and congestion control, current implementations and give some future directions and possibilities.
It’s 2019 and You are Still Getting Hacked by Otha Mabry (Cisco)
Learn emerging trends and techniques of open source research intelligence used to start a intrusion campaign. Develop a paradigm shift in our understanding on how cyber attackers monetized criminal activity. In addition, utilize the techniques of reconnaissance through the discovery phase of corporate assets to determine how vulnerable an environment would be to an attacker. Furthermore, without the Darknet/Darkweb an organization has multiple attack vectors of Public Accessible Information that can be found to launch an attack.
Steelcase – Deployment of Open Networking & SDN in a Large Enterprise by Scott Schafer (Steelcase)
Open networking and white box switching has been the domain of hyperscalers for the last number of years. There are however enterprises who have embarked upon the deployment of open networking at different levels of scale. In this talk we will cover the deployment of multiple data center leaf/spine networks, and a campus aggregation network using white box switches with an open source-based OS and a controllerless SDN fabric. This talk will cover the reasons for our decision, the challenges we had during deployment and the benefits we have been able to achieve with this successful deployment in a large enterprise. We will also cover an IoT use case that took advantage of network overlays.
Seamless BFD (S-BFD) – Scalable Path Failure Detection Mechanism by Nagendra Kumar Nainar (Cisco)
Network resiliency and availability highly depends on the ability of the network to detect any failure rapidly. While current deployments leverages BFD for this purpose, it comes with negotiation and scale challenges that makes it hard to position as the OAM protocol for the recent SDN and Cloud environment. This session describes Seamless BFD (S-BFD), a new and simplified mechanism for deploying BFD that is SDN ready and more scalable comparing to traditional BFD. This session describes the architecture, deployment use case and evaluation that compares the performance of S-BFD with the traditional BFD.
How to get started at IETF
Quick panel discussion how to get involved working with IETF from panel members who authored or participated in RFC writing process.