^ Top

NANOG 70 Agenda

All times listed in Pacific Time.

**Agenda is subject to change.**

The meeting will be Webcast

The NANOG 70 Agenda icalendar feed is now available! 
Download the .ics file and import into your calendar, or subscribe to the URL: 
http://www.nanog.org/sites/default/files/ical/nanog.ics 

Download the Guidebook app! If you already have the app, you can search for NANOG 70 starting June 1st and download the guide to your mobile device.

 

NANOG 70 Host Sponsor: NTT Communications
NANOG 70 Connectivity Sponsor: NTT Communications

Meeting Floor Plans: Lobby Level, Second Floor, Third Floor

 

Sunday, June 4 2017
Time/Webcast:Room:Topic/Abstract:Presenter/Sponsor:Presentation Files:
8:30am - 6:00pm

Cedar BallroomHackathonSpeakers:

  • L Sean Kennedy, NANOG Program Committee Chair
  • L Sean Kennedy is an active member of the Internet Engineering community and a Director of Network Engineering at XO Communications. He has been a member of the NANOG Program Committee since 2013.
  • TJ Trask.
Sponsors:
4:00pm - 6:00pmGrand Foyer NorthSunday, Meeting Registration
7:00pm - 11:00pmLucky Strike

Sunday Evening Social Event

Lucky Strike 700 Bellevue Way NE Suite 250 Bellevue, WA 98004 http://www.bowlluckystrike.com/locations/bellevue/

View full abstract page.
Sponsors:
Monday, June 5 2017
Time/Webcast:Room:Topic/Abstract:Presenter/Sponsor:Presentation Files:
7:30am - 9:45amEvergreen BallroomMonday Extended BreakfastSponsors:
8:30am - 5:00pmGrand Foyer NorthRegistration
10:00am - 10:30am

Grand BallroomConference OpeningSpeakers:

  • L Sean Kennedy, NANOG Program Committee Chair
  • L Sean Kennedy is an active member of the Internet Engineering community and a Director of Network Engineering at XO Communications. He has been a member of the NANOG Program Committee since 2013.

  • David Temkin, NANOG Board Chair
  • Dave Temkin is the Director of Global Networks for Netflix. Having been hired to build the Open Connect CDN, he is responsible for all network architecture and strategy as well as the operations fo the Netflix network (AS2906). Before Netflix, he was at Yahoo!, where he focused on Layer 4-7 network architecture, having been brought in through the successful acquisition of Right Media where he was the Global Head of Networks. In his spare time he enjoys travel and philanthropy - both through volunteering at technical organizations such as NANOG, where he is vice chair, Open-IX, where he is the chairman and co-founder and FL-IX, where he is also chairman and cofounder, as well as on the board of Children of Bellevue.

  • Michael Wheeler, NTT Communications
  • Executive Vice President NTT Communications Global IP Network Michael Wheeler is the executive vice president of the NTT Communications Global IP Network at NTT America, a subsidiary of NTT Communications. Mr. Wheeler is responsible for all financial, engineering, operational and customer relationship activities for the Global IP Network business unit, known as GIN. Prior to his current role, Mr. Wheeler led the Global IP Network sales and business development organization, including direct sales, account management, indirect channel sales, sales engineering, marketing, product management and business development activities. Mr. Wheeler was part of the original start up team with Verio in 1997, where he led several different sales organizations. He transitioned to a leadership role at NTT America following the acquisition of Verio by NTT America’s parent company, NTT Communications, in August of 2000.
10:30am - 10:45am

Grand BallroomHackathon ReadoutSpeakers:

  • L Sean Kennedy, NANOG Program Committee Chair
  • L Sean Kennedy is an active member of the Internet Engineering community and a Director of Network Engineering at XO Communications. He has been a member of the NANOG Program Committee since 2013.
  • TJ Trask.
10:45am - 11:30am

Grand Ballroom

Keynote: Revisiting Statistical Multiplexing in Your Network

A key component to any network is the ability to aggregate traffic. In this regard, statistical multiplexing is not a new concept; it was used in the 1960s by the Bells. Making the most efficient use of the bandwidth available is a key component of network designs. Inefficiencies cost time, money, and waste resources. Len will discuss how to leverage statistical multiplexing in your network.

View full abstract page.
Speakers:

  • Len Bosack, XKL
  • From pioneering the commercialization of router technology as the co-founder of Cisco Systems in the 80s to creating forward-thinking optical networking innovations today, Len has been a key contributor to the networking industry for decades.
11:30am - 12:00pm

Grand Ballroom

Seattle Internet Exchange

The Seattle Internet Exchange started in 1996 with a traceroute between devices on the 19th floor of the Westin Building carrier hotel that took 111ms RTT and began its progression into a community Internet exchange point launched in 1997. This presentation introduces the SIX and recounts how it surmounted the technical, business, and organizational challenges it faced. Where does the SIX stand after 20 years and what could that mean for other community IXPs?

View full abstract page.
Speakers:
  • Chris Caputo.
  • Nikos Mouat.
  • Jared Reimer.
12:00pm - 1:15pmEvergreen Ballroom (E-I)Monday Welcome LunchSponsors:
12:00pm - 1:15pmEvergreen Ballroom (A-D)Newcomers Lunch (Invite Only)Sponsors:
1:30pm - 2:15pm

Grand Ballroom

DDoS Trends in Last 20 Years

There seems to be an alarming amount of 'new' DoS and DDoS events that exasperate the feeling of apathy and hopelessness to combat these attacks. This talk will walk thru the history of DDoS trends in the last 20 years and show how many of the techniques are just variants of SOS and how network operators and developers can collectively work together to deter and protect themselves against current DDoS attacks.

View full abstract page.
Speakers:

  • Merike Kaeo, Farsight Security
  • Ms. Kaeo has over 25 years experience leading strategic technology direction for startups, corporate security initiatives and evolving technologies. Ms. Kaeo is the founder and CEO of Double Shot Security and has in the past had positions as the Chief Information Security Officer at IID, the CTO of Guardtime, and has served on numerous technical advisory boards. She led the first security initiative for Cisco Systems in the mid 1990s and authored the first Cisco book on security - translated into more than 8 languages and leveraged for prominent security accredidation programs such as CISSP. She has contributed to numerous IETF standards, is a member of the ICANN Security and Stability Advisory Council (SSAC) and the FCC's Communictaions Security, Reliability and Interoperability Council (CSRIC). She earned a MSEE from George Washington University and a BSEE from Rutgers University.
2:15pm - 2:45pm

Grand Ballroom

When BGP meets Big-data

The Internet is very much ‘alive’, with millions of BGP events occurring every day. BGP monitoring and reporting tools have been in use for many years now and are a standard part of many operators toolkits. With all this 'noise', how do we extract and understand the 'signal'? Can we apply techniques from other domains in this pursuit? It is well understood that systems can produce more data than we know what to do with. Asking questions of the data, using analytics techniques will help to open up new operational and business insights. In this presentation, we propose that big-data analytics should be a foundational element of the next generation of operational analysis systems. Applications range from Data-center performance and operations, extracting insights from data-plane traffic through to Smartcities and IoT-centric applications. This presentation will show how the PNDA.io platform can be used to realise such insights and present use cases that demonstrate some of the possibilities enabled by the use of big data analytics. As an illustration, we will show the use of the PNDA.io platform in combination with OpenBMP, in order to be able to ask questions about the stability of Internet, the consistency of paths through the Internet, the frequency of changes and how such behaviours are changing over time. PNDA.io is a Linux Foundation Collaborative Project.

View full abstract page.
Speakers:

  • Tim Evens, Cisco
  • Tim Evens has over 19 years of experience as a network engineer and programmer in wide range of industries including carrier and internet service providers, financial trading, healthcare, retail and technology. Tim has been active as a double CCIE for 14 years. In his current position, Tim works in the Chief Technology and Architecture Office at Cisco and is responsible for establishing inter-networking data analytics. He is the maintainer of OpenBMP (www.openbmp.org).
2:45pm - 3:00pm

Grand Ballroom Hackathon runner upSpeakers:

  • L Sean Kennedy, NANOG Program Committee Chair
  • L Sean Kennedy is an active member of the Internet Engineering community and a Director of Network Engineering at XO Communications. He has been a member of the NANOG Program Committee since 2013.
3:00pm - 3:30pmGrand Foyer Monday PM BreakSponsors:
3:30pm - 4:00pm

Grand Ballroom

Bootstrapping IX-Denver: A Cooperative IX Case Study

What does it take to build an IX? How could we as an industry make this process easier? Through the narrative of our first year at IX-Denver, this talk will answer the why, where, who, and how of bringing a new cooperative IX online. From hardware to software, location to community, motivation to execution, you will walk away with a solid understanding of how we built IX-Denver, and how you might do the same. If you've ever thought about starting a local or regional IXP, or if you simply want to know what's involved - come learn from our bumps and bruises!

View full abstract page.
Speakers:

  • Chris Grundemann, Myriad Supply
  • Chris Grundemann is a passionate Internet Technologist and a strong believer in the Internet's power to aid in the betterment of humankind. In his current role as Principal Architect at Myriad Supply he is expressing that passion by helping clients build bigger, faster, more efficient networks that are easier to operate and scale. Chris has well over a decade of experience as both a network engineer and architect designing, building, and operating large IP, Ethernet, and Wireless Ethernet networks. Chris holds 8 patents in network technology and is the author of two books, as well as several blogs, and various other industry papers. As a volunteer he is currently serving as President of IX-Denver, Program Committee (PC) member for AfPIF, ISOC-NY liaison to NYNOG, and Chair of the Open-IX standards development committee. Chris has held previous positions with Markley Group, Internet Society, CableLabs, tw telecom, CO ISOC, ARIN, NANOG, CEA, UPnP, DLNA, RMv6TF, and several others.
3:30pm - 4:45pm

Regency Ballroom

Tutorial: Segment Routing

Segment Routing (SR) allows network operators to direct traffic from source to destination through an ordered set of segments. The SR path may be identical to or different from the IGP least-cost path. SR differs from traditional traffic engineering (TE) mechanisms in that it moves TE state from transit routers into the packet header. This tutorial addresses the SR forwarding as it is realized for MPLS and IPv6. It also addresses OSPF, ISIS and BGP extensions for SR by Richard Turkbergen and Ron Bonica

View full abstract page.
Speakers:

  • Ron Bonica, Juniper Networks
  • Ron Bonica currently serves as Director of Juniper Networks Routing Protocol Daemon Architecture Group. He has co-authored fifteen IETF RFCs and co-chairs the IETF V6OPS Working Group. In the past, Ron has served as co-director of the IETF Operations and Management Area. Prior to joining Juniper Networks, Ron was senior manager of engineering for a major Internet Service Provider.
4:00pm - 4:45pm

Grand Ballroom

Interconnection Track: PeeringDB Update

PeeringDB is an indispensable resource for peering information, and is now required by many networks in order to peer. In this presentation we’ll talk about the new 2017 strategic plan, goals and roadmap objectives for this year.

View full abstract page.
Speakers:

  • Aaron Hughes, 6connect
  • Aaron Hughes is the President and CEO of 6connect, a leader in network provisioning automation including global resource management of IPv6/IPv4 Addressing, DNS/DNSSEC Control from the cloud or behind the firewall for Service Providers and Enterprises. He is deeply rooted in the community, supporting organizations such as ARIN, NANOG, RIPE, Rocky Mountain IPv6 Task Force, go6.si and GPF. He is also a contributor to IFG, EPF, Interop, Cable Labs, EuroIX, BCP/BCOP and IPv6 outreach programs. Aaron has held network and system architecture and Sr. level management roles at Lockheed Martin, Cariden Technologies, Terremark, Certainty Solutions, UnitedLayer, Quest Technologies, RCN, UltraNet and Channel(1) Communications among others. He is the President and Chairman of the Board for PeeringDB.
5:00pm - 6:30pmCedar BallroomPeering Coordination ForumSponsors:
7:00pm - 11:00pmParlor Bellevue

Monday Evening Social Event

Parlor Live 700 Bellevue Way NE Suite 300 Bellevue, WA 98004 http://www.parlorlive.com

View full abstract page.
Sponsors:
Tuesday, June 6 2017
Time/Webcast:Room:Topic/Abstract:Presenter/Sponsor:Presentation Files:
7:30am - 9:45amEvergreen BallroomTuesday, Extended BreakfastSponsors:
8:30am - 5:00pmGrand Foyer NorthRegistration
9:00am - 10:00amEvergreen ABCDMembers Breakfast (Invite Only)
10:00am - 11:00am

Grand BallroomPanel Discussion: Finding your next gigSpeakers:
  • Mehmet Akcin, Yahoo Inc.
  • Michael Ballenger.
  • Jeff DeClue.
  • Peter Hartzell.
  • Peter Hoose, Facebook, Inc.
  • Peter has spent the last fifteen years attempting to automate himself out of a job, thus far unsuccessfully. Rest assured, he’ll keep at it until the job is done. In his current role as Network Infrastructure Manager at Facebook, his teams combine the power of automation with solid network engineering to keep facebook’s global backbone, and datacenter networks running fast, reliably, and efficiently. Today, they are building the infrastructure to deploy wedge, six-pack and fboss, facebook’s open switching platform and network operating system. Prior to this, Peter worked as a Network Engineer at Facebook, and further in the past as Senior Network Engineer and Architect at NTT America where he built custom dedicated hosting solutions for customers large, and small.
  • Dan Seifert.
11:00am - 11:30am

Grand Ballroom

The State of Open Source Routers

Open source routers are an interesting alternative to many vendor driven options, but the market is immature, and the offerings slim. This session will walk through several of the options available for open source and disaggregated routing platforms currently available, as well as a quick overview of some architectural bits and pieces. This overview will include information on FR Routing, a new Quagga fork that was recently taken under the Linux Foundation.

View full abstract page.
Speakers:

  • Russ White, Ericsson
  • Russ White has scribbled a basket of books, penned a plethora of patents, written a raft of RFCs, taught a trencher of classes, nibbled and noodled at a lot of networks, and done a lot of other stuff you either already know about or don't really care about. You can find Russ at 'net Work, the Internet Protocol Journal, PAcket Pushers, LinkedIn, and his author page on Amazon.
11:30am - 12:00pm

Grand Ballroom

Turning the Network: Lessons learned from redeploying an ISP's network

This talk will go through the process Cox Business went through to upgrade our internal network. At the beginning of the project, the network consisted of a number of separate islands, with separate OSPF area 0s, different metrics, different BGP policies, and utilizing basic point-to-point pseudowires between specially configured nodes to manage cross-market functionality. The end product is a cohesive nationwide multi-instance multi-topology IS-IS domain, with IPv4 and IPv6 routing capabilities, consistent BGP topologies supporting L2VPNs, L3VPNs, and SDN, and a defense-in-depth security model applied. This talk will go through the phases of the project, from initial concept to our final deployments, with focuses on not only the technology deployed, but also a lot of the "lessons learned" from the work. The different phases included initial project scoping, an initial high level design, consulting with other engineers both internally and externally, and numerous revisions to the HLD, low-level design, testing, and deployment. The large amount of time spent on the HLD and corresponding low-level design and testing resulted in a deployment that had no impact to end customers, and only one minor issue that arose during the work. The presentation is aiming to talk more about the "whys", as opposed to straight configuration examples. While some configurations will be shown, they are presented in the context of highlight where issues were buried, or other neat tricks utilized to work around issues. The final section will include both "Why We Did" and "Why We Didn't" sections, diving more into the thinking that led to the decisions made.

View full abstract page.
Speakers:
  • Andrew Gray.
12:00pm - 1:15pmEvergreen BallroomTuesday LunchSponsors:
12:00pm - 1:15pmEvergreen ABCWomen In Technology Lunch Speakers:
  • H. Jezzibell Gilmore.
  • Merike Kaeo, Farsight Security
  • Ms. Kaeo has over 25 years experience leading strategic technology direction for startups, corporate security initiatives and evolving technologies. Ms. Kaeo is the founder and CEO of Double Shot Security and has in the past had positions as the Chief Information Security Officer at IID, the CTO of Guardtime, and has served on numerous technical advisory boards. She led the first security initiative for Cisco Systems in the mid 1990s and authored the first Cisco book on security - translated into more than 8 languages and leveraged for prominent security accredidation programs such as CISSP. She has contributed to numerous IETF standards, is a member of the ICANN Security and Stability Advisory Council (SSAC) and the FCC's Communictaions Security, Reliability and Interoperability Council (CSRIC). She earned a MSEE from George Washington University and a BSEE from Rutgers University.
1:45pm - 2:15pm

Grand Ballroom

The Internet is Flat: Revisited

IIn 2011 at NANOG52, a small Tier3 ISP AS19653 joined NANOG. Also in 2011, this small ISP read the paper - “The Internet is Flat: Modeling the Transition from a Transit Hierarchy to a Peering Mesh” The forecasts in this paper were used to guide business and network planning. Actual network data was collected from AS19653 from 2010 to present. This data illustrates the predictions in the paper “The Internet is Flat” and shows a graphic history about the “Transition from a Transit Hierarchy to a Peering Mesh”. Part 1 – Overview of the 2010 paper “The Internet is Flat” (CIADA) Part 2 – Presentation of CTS Telecom’s transition for all transit to over 85% peering (CTS). NANOG and its role in CTS developing peering relationships and knowledge.

View full abstract page.
Speakers:

  • Robert Bender, CTS Telecom
  • Robert Bender is Network Operations Manager at CTS Telecom, a 105-year old independent service provider in SW Michigan. In 18 years at CTS Telecom, Robert has lead Tier 2 ISP operations, softswitch deployment, and currently is the lead optical and IP networking architect. His vision continues to move CTS Telecom forward, from legacy systems to Carrier Ethernet and optical services.
2:15pm - 3:00pm

Regency Ballroom

Building a nationwide carrier for <$1M - war stories & lessons learned

From 2006 to 2010, Jonny and Tim were part of a small engineering team that built New Zealand’s 3rd nationwide telecommunications network, laying multiple thousands of kilometers of fibre and lighting up a DWDM transmission and MPLS telco network. They did this with 58 POPs on under $1M of routing equipment, while scaling to take a significant proportion of the internet traffic in the country. Requiring a fairly large amount of “creative engineering” and some Kiwi ingenuity, they’ll talk about how they managed to pull this off, some of the lessons learned, pitfalls, and hilarious stories involved in this achievement.

View full abstract page.
Speakers:

  • Tim Hoffman, Twitter
  • Following a decade of building telecommunications networks in New Zealand, Tim is lead engineer for Twitter's global backbone.
  • Jonny Martin, Limelight Networks.
3:00pm - 3:30pmGrand Foyer Tuesday PM BreakSponsors:
3:30pm - 4:00pm

Grand BallroomLightning Talks
4:00pm - 5:30pm

Grand Ballroom

Tutorial: Everything You Always Wanted to Know About Optical Networking

This popular tutorial tailored for Network Engineers has been updated to cover the latest technologies.

View full abstract page.
Speakers:

  • Richard A Steenbergen, PacketFabric
  • Richard Steenbergen is co-founder and Chief Technology Officer of PacketFabric, a next-generation Network-as-a-Service interconnection and transport solutions provider. Prior to starting PacketFabric, Richard served as CTO of GTT Communications, a leading global IP/MPLS backbone and Tier 1 network operator in over 80 countries. Richard was also the founder and CTO of nLayer Communications for 11 years, served as a Sr Network Architect for various large NSPs, and served as a Sr Software Engineer helping developed advanced optimized routing techniques. Richard has been attending NANOG since 1999, and is a frequent speaker and contributor.
4:15pm - 4:45pm

Regency Ballroom

Hashing on broken assumptions

Widely used techniques such ECMP and LAG rely on hardware-based hashing performed on a packet five-tuple to uniformly spread traffic over different links while ensuring that packets belonging to the same flow take the same path to avoid packet reordering. These techniques operate under the assumption that the underlying hashing algorithm is correctly implemented and configured and uniformly spread traffic. But is it really the case? This talk gathers years of operational experience from the perspective of a CDN on hashing in the wild gone horribly wrong - from hardware limitations, to buggy software across both switches and middleboxes. Understanding these broken assumptions is critical when scaling load balancing functions at the network layer, as well as operating anycast networks.

View full abstract page.
Speakers:

  • Lorenzo Saino, Fastly
  • Lorenzo Saino is a software engineer at Fastly, where he works on network systems and load balancing problems. Prior to that, he was a PhD student at University College London. His thesis investigated design issues in networked caching systems and he was awarded the Fabrizio Lombardi prize in 2016 for his research.
4:45pm - 5:15pm

Regency Ballroom

BGP Optimal Route Reflection - Technology and Deployment

Optimal Route Reflection is a software enhancement to the BGP best-path selection algorithm to eliminate the need for RRs be geographically close to their client. This enhancement along with virtualized RRs provides tremendous flexibility in how operators can deploy RRs today. This general session talk will give an overview of why ORR has come to be, its basic operation, current vendor landscape, deployment options, and how the technology used in ORR may be extended in the future to provide even greater control over the BGP path selection process.

View full abstract page.
Speakers:

  • Phil Bedard, Cisco
  • I began working at my first ISP at the age of 16, working tech support, building servers, and managing a network of Proteon routers. I continued to work at service providers for over 20 years, most recently as a Principal Engineer at Cox Communications, a position I held for ~9 years. I joined Cisco in March 2017 as a Technical Marketing Engineer in the SP Networking Software and Automation group, focused on network architecture and helping build automation products for the operator community. I have resided in Atlanta, GA for 10 years, originally hailing from SE Wisconsin. I have a B.S. in Computer Science from the University of Wisconsin-Parkside. In my spare time I enjoy racing in amateur motorsports, backcountry hiking, and tinkering with just about everything.
5:15pm - 5:45pm

Regency Ballroom

Lost in Fat Tree forest and route out

Clos and Fat-Tree have gained prominence in today's networking, primarily as a result of a the paradigm shift towards a centralized data-center based architecture that is poised to deliver a majority of computation and storage services in the future. The existing set of dynamic routing protocols was geared originally towards a network with an irregular topology and low degree of connectivity and consequently several attempts to adapt those have been made in order to enable CLOS/FatTree Fabrics. In looking at the problem through the very lens of its requirements an optimal approach does not seem to be a simple modification of either a link-state (distributed computation) or distance-vector (diffused computation) approach but rather a mixture of both, colloquially best described as 'link-state towards the spine' and 'distance vector towards the leafs'. This session will describe prionciple of novel routing protocol - RIFT (draft-przygienda-rift-01) - targeted for CLOS/FatTree topologies. The techical motivation for new protocol as well techniques used to overcome limitation of BGP and ISIS/OSPF would be discussed.

View full abstract page.
Speakers:

  • Rafal Szarecki, Juniper Networks
  • Rafal Szarecki (JNCIE 136) has 20 years of experience in networking industry. He has started his career working for ISP in Poland as Network Operation Centre stuff member in 1996. Later in his career he work in Customer Support consultant and then as Network Architect in Ericsson. In 2006 Rafal joined Juniper Networks Professional Services Team and later become Principal Consultant. He has worked on over 50 different projects, with more then 20 Telecoms and ISP including some of global reach such British Telecom, Telefonica, Telenor. He designed and/or helped in deployment over 20 networks in multiple countries in Europe and Middle East. During this time he was also active member of PLNOG. Currently Rafal held position of Solution Architect in Routing Product Team in Juniper Networks in Sunnyvale, CA.
6:00pm - 8:00pmEvergreen BallroomBeer n GearSponsors:
8:00pm - 12:00amLucky Strike

Tuesday Social Event

Lucky Strike 700 Bellevue Way NE Suite 250 Bellevue, WA 98004 http://www.bowlluckystrike.com/locations/bellevue/

View full abstract page.
Speakers:
  • Darrieux Harvey, NANOG.
Sponsors:
Wednesday, June 7 2017
Time/Webcast:Room:Topic/Abstract:Presenter/Sponsor:Presentation Files:
7:30am - 9:45amEvergreen BallroomWednesday, Extended BreakfastSponsors:
8:30am - 5:00pmGrand Foyer NorthRegistration
10:00am - 11:00am

Grand Ballroom

The Networking Grand Challenge

This talk proposes a Networking Grand Challenge: to build a self-driving network – one that runs itself in all aspects, from self-discovery to configuration to monitoring to fault detection and remediation to real-time optimization. While there are projects to improve network management via automation, zero-touch provisioning, self-organizing networks and the like, this proposal attempts to encompass and leapfrog these efforts to get to the end goal: full autonomy. It suggests some building block technologies needed to achieve this; however, this is not expected to be a six-week or six-month effort. It also takes a stab at some of the implications of achieving full autonomy, and suggests augmentation as an alternative to autonomy. The goal is to lay out a vision for network management of the future, and to start an open discussion on what it would take from a technology point of view, and what outcomes to expect.

View full abstract page.
Speakers:

  • Kireeti Kompella, Juniper Networks
  • Dr. Kireeti Kompella is Senior Vice President and Chief Technology Officer (CTO) for Juniper's engineering team. Before taking on his current role, he served as CTO, SDN at Juniper Networks. Previously, Kompella was CTO at Contrail Systems, which was acquired by Juniper in December 2012. Prior to joining Contrail, he was CTO and Chief Architect, Junos at Juniper Networks.
11:00am - 11:30am

Grand Ballroom

SONATA: Scalable Streaming Analytics for Network Telemetry

Current solutions for network telemetry are ill-suited for security or performance troubleshooting as they offer limited expressibility and are harder to scale as the number of monitoring queries or the volume of data increases. Expressing queries as dataflow operations (i.e., map/ reduce/ distinct/ group, etc.) over packet tuples addresses the expressibility problem, but scaling such a system is non-trivial. We observe that network telemetry can benefit from two key observations: (1)~underlying data plane can process the packets at line rate, and (2)~small portion of the total traffic satisfies the query for most network monitoring applications. This talk presents the design and implementation of SONATA, a stream-based network telemetry system that allows an operator to express network-wide queries as dataflow operations over packet tuples. Given a query, SONATA automatically determines the optimal plan, (1)~refining the input query to iteratively zoom in over portions of traffic that satisfy the query, and (2)~partitioning the refined queries across the network switches and the stream processing system. We implement several example queries that are motivated by real-world security and troubleshooting scenarios and quantify the scalability benefits of SONATA for these queries using traffic traces from different production networks.

View full abstract page.
Speakers:

  • Arpit Gupta, Princeton University
  • Arpit Gupta is currently a fourth year Ph.D. student in CS at Princeton University. At Princeton, he works under the supervision of Dr. Nick Feamster and is also mentored by Dr. Jennifer Rexford. His research focuses on the intersection of Internet Routing, Software Defined Networks (SDN), and Big Data. He is one of the creators of software-defined IXPs (SDXs). Before joining Princeton, he completed his Masters from NC State University and undergrad from Indian Institute of Technology, Roorkee, India. Web: http://www.cs.princeton.edu/~arpitg/
11:30am - 12:00pm

Grand Ballroom

Network Telemetry at Yahoo!

Providing 1 billion monthly active users with responsive, rich applications requires a large scale network. Locked within processes running on network devices are valuable bits of control and data plane metrics like prefix usage, peer interface utilization and routing session flaps. By making this data available to any number of subscribers, we enable Yahoo! Engineers to create cost saving data visualizations and anomaly detection software. This paper explains the challenges encountered and architecture decisions made in building our real time network telemetry stack that currently polls millions of metrics from dozens of sites on five continents. A key goal of our system is to minimize the effort required to poll a new device type or write a new consumer application. To accomplish this, we abstracted scale away from engineers looking to poll devices and consumption away from engineers looking to build consumer applications. Our Python polling layer is built to be future proof, modular and horizontally scalable. We chose to use Python as our language because of its readability and community support. Python’s open source community provides a ready made plugin system called Yapsy. Polling plugins in our system are Yapsy plugins that specify how to get and clean data from a device before placing the results onto a Kafka bus. The platform then horizontally scales (unlike MRTG or Cacti) by scheduling the plugin through Celery, a Python distributed task queue. This yields many benefits, including the freedom to use the best polling method for a given device and the luxury of not needing to worry about scaling your plugin. For instance, where vendors support a robust API, we use that, but for API deficient vendors, we poll by SNMP instead. We also developed configuration driven SNMP polling that allows us to define SNMP table relations in configuration rather than code. This approach eases the mental burden of cross-SNMP table correlations, and allows us to poll new metrics without having to touch source code.

View full abstract page.
Speakers:

  • Matt Hudgins, Yahoo!
  • Matt Hudgins is currently a Senior Software Engineer at Yahoo! building scalable network analysis and optimization applications using open source software. He was previously a Software Engineer at Cisco Systems where he developed network operating systems for service provider networks. His open source contributions range from frontend tutorials to network monitoring tools.
  • Varun Varma.
12:00pm - 1:15pm Wednesday Lunch (On Your Own)
1:30pm - 2:00pm

Grand Ballroom

Network Automation And Programmability: Reality Versus The Vendor Hype When Considering Legacy And NFV Networks

This talk will focus on: 1 - The journey involved in network automation, at a high level 2 - Programmability as an approach to automation 3 - Legacy and Future Automation approaches 4 - Vendor Hype: NETCONF/YANG and NFV 5 - The reality we face today 6 - An approach to future network automation Programmability is the ability to manage network devices in a similar manner to the way you can manage software and servers in an IT environment today. We are gradually seeing the emergence of network devices that present APIs for use in managing them, such as the Cisco Nexus devices. This is opening up new automation possibilities. There is an architecture, and open source tools such as NAPALM, that make this more of a reality as time progresses, as well. This architecture/toolset allows programmable automation to be applied to legacy equipment that isn't otherwise capable. Lastly, there is a common misconception that NFV is going to automatically be programmable since it is virtualized, but most VNFs out there today are just revisions of the same software from the physical device placed on a virtual machine. The result is a virtual device with all of the same limitations, in many cases MORE, than the physical device itself had. To make NFV programmable in the near term will require the same architecture as discussed for legacy network devices.

View full abstract page.
Speakers:

  • Patrick Moore, Itential, LLC
  • Patrick has been in the Telecommunications and Information Technology fields for 20+ years. Beginning in 2000 he took on a leadership role in the implementation of networks for voice and data for a CLEC in the Southeast. Prior to that he held various technician roles for both wireless and wireline carriers maintaining networks in the field and at switch sites. Patrick transitioned to a vendor role with Alcatel-Lucent as his career progressed, managing teams of analysts, architects, and developers responsible for OSS solutions across the Americas Region. In the last few years at Itential he has been involved in the emergence of SDN and NFV by assisting clients with the evolution of their networks to the new paradigm of a programmable network, leveraging both vendor and open source solutions.
2:00pm - 2:30pm

Grand Ballroom

Open Undersea Cable Systems

Undersea bandwidth demand continues to increase for large-scale cloud providers, driven by phenomenal growth of cloud services and the globally distributed nature of data centers. To efficiently meet this demand, cloud providers have driven the open cable system (OCS) concept. In this talk, we discuss Microsoft’s approach to undersea cable systems and the evolution from complex closed systems to simple, efficient open cable systems.

View full abstract page.
Speakers:

  • Jamie Gaudette, Microsoft
  • Jamie Gaudette is the Manager of Optical Networking at Microsoft in Redmond, WA, USA, where he designs and builds open data center networks, and develops SDN software. Before Microsoft, Jamie designed industry leading networking hardware and software with Ciena and Nortel in Ottawa, Canada. He has 4 US patents in optical networking, is an author of the current ITU G.977 specification for undersea cables, and has a Master’s degree in Electrical Engineering from Queen’s University in Canada.
2:30pm - 3:00pmGeneral SessionTBA
3:00pm - 3:30pm Wednesday BreakSponsors:
3:30pm - 4:00pm

Grand BallroomLightning Talks
4:00pm - 4:45pm

Grand Ballroom

Designing Multi-tenant Data Centers using EVPN Integrated Routing and Bridging (IRB)

EVPN-IRB (Integrated Routing and Bridging) is a technology that leverages BGP EVPN as common overlay control plane to enable VPN routing and bridging service over an MPLS or IP underlay fabric. Point to multi-point bridging service enables VLANs to be stretched across data center IP or MPLS fabric, while VPN routing service enables inter-subnet routing across these stretched subnets. It hence allows for flexible workloads with seamless VM mobility across the stretched subnet. When designing a multi-tenant data center overlay solution using BGP EVPN control plane, multiple design options exist with respect to supporting this IRB service in the overlay: o Symmetric EVPN-IRB: with distributed first-hop any-cast GW on the ToRs o Asymmetric EVPN-IRB: with distributed first-hop any-cast GW on the ToRs o Centralized EVPN-IRB: with centralized first-hop any-cast GW on the border leafs OR DCI / DC Edge routers This talk will go over key architectural aspects that differentiate each of the above EVPN-IRB designs and focus in-depth on Symmetric EVPN-IRB design. For symmetric EVPN-IRB design, it will cover: o Overlay network architecture with respect to distributed any-cast GW and Multi-homing using EVPN-LAG o Relevant EVPN control plane constructs for IP+MAC reachability, fast convergence, BUM traffic forwarding with Designated Forwarder (DF) election and Split Horizon Group (SHG) filtering o End to End packet walk-thrus for intra-subnet bridged BUM and known unicast traffic o End to End packet walk-thrus for inter-subnet routed unicast traffic, including ARP resolution across distributed any-cast GW. o VM Mobility Neeraj Malhotra Principal Engineer, Cisco [email protected] Ahmed Abeer Technical Marketing Engineer, Cisco [email protected]

View full abstract page.
Speakers:
  • Ahmed Abeer.
  • Neeraj Malhotra, Cisco
  • Principal Engineer Cisco San Jose, CA 95134
4:45pm - 5:15pm

Grand Ballroom

Status and Solutions for Whois Data Accuracy

Internet operability, public safety (law enforcement), and the ability to prevent the hijacking of address space are just a few things impacted by the quality of Whois data. The need to improve the accuracy of this Whois registration data is increasing with the depletion of the IPv4 free pool, and the increasing number of registration transfers. A breakdown of the Whois data will be examined with a look at points of contact, registered organizations, and orphaned records. Transfer statistics will also be provided including completion rates, average process times, and known issues. What ARIN is doing today to obtain and maintain accurate data will also be reviewed with a high level look at current policies, business practices, and contractual requirements. This will be accompanied by a look at ARIN policy changes currently being considered, along with an open discussion on ways to improve data accuracy, update incentives, and real world impacts of Whois data on network operators.

View full abstract page.
Speakers:
  • Dan Alexander.
  • Leslie Nobile
  • Leslie Nobile is the Senior Director of Global Registry Knowledge at the American Registry for Internet Numbers (ARIN). Reporting directly to the CEO, her main role is to research and recommend Registry improvements both within ARIN and across the Registry system, with a focus on data integrity and quality. Prior to that, Leslie served as ARIN’s Director of Registration Services for 14 and ½ years where she was responsible for the delivery and execution of ARIN’s core function of allocating and registering Internet Number Resources. Leslie has over 25 years of experience in the Internet industry, where she has held various technical management positions. Her previous work includes Internet registry operations at the DDN/DoD Network Information Center (NIC), as well as technical support of the development and expansion of the Defense Data Network (DDN), a high-speed military data network that evolved from the ARPANET. Leslie received her B.A. from the American University in Washington, D.C.
5:15pm - 5:30pm

Grand Ballroom

Proactive Network Bug Finding with Minesweeper

Over the past 4 decades, networks have become increasingly complex as scalability, quality of service, robustness, and fault-tolerance requirements have grown to meet the demands of the 21st century. Much of the complexity of networks lies in their configuration. Configuration files often have thousands of lines of assembly-like directives that control a huge variety of parameters for various protocols and their interactions. The combination of the size and low-level nature of configurations make it hard for humans to reason about individual device behaviors and even harder to reason about the network behavior that emerges through their interactions. To make matters even worse, the possibility for network failures makes this reasoning even more challenging. It is not surprising then, that configuration errors that lead to costly outages are all-too-common. Systematic surveys of network outages routinely point to configuration error as one of the biggest contributors, and configuration-induced outages at major networks routinely make the news ever few months. In this talk, I will present Minesweeper, a new general-purpose network analysis tool that can check unmodified network configurations against a wide variety of desirable properties such as reachability between devices, equal or bounded path lengths for multiple devices, symmetric load balancing, local router equivalence, and others. Prior tools are either based on heuristics (e.g., rcc) or model network behavior under a single set of external routing announcements (e.g., Batfish). In contrast, any operator-specified property checked by Minesweeper is checked against all possible external environments (i.e., collections of eBGP advertisements from peers), and for all possible k network failures, where k is specified by the user. We have evaluated Minesweeper on a collection of real configurations, showing that it is effective at finding issues and that it is scalable, often checking rich properties of large networks in minutes.

View full abstract page.
Speakers:

  • Ryan Beckett, Princeton University
  • I am a 4th year PhD student at Princeton studying programming language theory, design and implementation with my advisor David Walker. I received Bachelors degrees in Computer Science and Mathematics from the University of Virginia. Recently, I have been working on applications of synthesis, verification, and domain-specific languages to network configuration.
5:30pm - 5:45pm

Grand Ballroom

Simplifying Network Configuration with Propane

Building reliable computer networks is becoming increasingly important as more and more critical services move to the cloud and new devices come online. Yet despite its importance, designing and configuring networks correctly remains a highly manual and error-prone process. Unfortunately, even small mistakes when configuring network devices can result in widespread and costly disruptions in service. One of the fundamental reasons for the prevalence of misconfigurations is the mental mismatch between the intended high-level policies and requirements that operators have in mind when designing the network, and the low-level mechanisms provided by today's vendor-specific configuration languages. Network Requirements often involve network-wide properties such as keeping traffic within a geographic region, but configurations describe the low-level behavior of individual devices. To complicate matters, ensuring correct behavior of the network becomes even more challenging when operators must consider the possibility of network failures. In this talk I will present Propane, a new high-level language for simplifying network configuration. The language allows operators to specify objectives using high-level constraints on both the shape and relative preference of different traffic paths. Given a Propane policy and a network topology, the Propane compiler generates router-level BGP configurations for every device in the network that can run on unmodified vendor hardware. The compiler guarantees that the compiled configurations will continue to correctly implement the high-level policy under all possible combinations of failures without any centralized coordination. We evaluated Propane by writing policies for real configurations taken from data center and backbone networks from a large cloud provider. For each network, the corresponding Propane policy required less than 50 lines of code, and even for the networks with thousands of routers, the compiler was able to generate configurations in under 9 minutes.

View full abstract page.
Speakers:

  • Ryan Beckett, Princeton University
  • I am a 4th year PhD student at Princeton studying programming language theory, design and implementation with my advisor David Walker. I received Bachelors degrees in Computer Science and Mathematics from the University of Virginia. Recently, I have been working on applications of synthesis, verification, and domain-specific languages to network configuration.
5:45pm - 6:00pm

Grand BallroomConference ClosingSpeakers:

  • Betty Burke, NANOG
  • Currently serving as the NANOG Executive Director, responsible for all aspects of NANOG, reporting to the Board of Directors. Previous 37 years of experience serving in technology, business, and management within the Michigan Information Technology Services, University of Michigan, and Merit Network. Proven leadership and experience in development of strategic and operational plans, creation and implementation of marketing campaign for conference center and high tech facilities including a data center, conference and office building, library and campus fiber assets. Proven operational success through project management, along with leadership through community and team building.

 

Thank you to the NANOG Service Sponsors: Cloudflare, DynServer Central

^ Back to Top