Full Abstract

Development and deployment of provider-provisioned VPNs is the latest VPN concept, having emerged during the past year. This panel will present the latest development of PPVPN technologies and models, and discuss ISP deployment status and experiences.

Speakers
Moderators - Jessica Yurcik, Cosine.
Panelist - Luyuan Fang, AT&T
Panelist - Kireeti Kompella, Juniper Networks
Panelist - Alvaro Retana, Cisco Systems

Full Abstract

Atoga Systems

Full Abstract

Speakers
Dave Plonka, University of Wisconsin - Madison

Full Abstract

The IETF AAA Working Group is in the final stages of defining DIAMETER as a potential standard protocol to support network access applications, focusing on dial and mobile access. There are also a number of applications that are now using or planning to use COPS (Common Open Policy Service) for provisioning network services such as QoS. This BOF will include:

• a brief presentation on the status of work at IETF on DIAMETER, COPS, and policy, and in the AAAARCH (AAA Architecture) IRTF research group.
  
  
• a brief presentation on possible transition models from RADIUS to DIAMETER in a "proxy" environment
  
  
• a brief discussion of combining services supported by RADIUS or DIAMETER with services supported by COPS, such as edge router access and provisioning
  
  
• open discussion of use of DIAMETER and/or COPS in current and future service offerings

Speakers
John Vollbrecht, Interlink

Full Abstract

What do you require your customers and peers to do, and not do? Can the spam? Forge source addresses? Inject routes they don't own? If not, why not? What else? Come by and talk with your, um, peers about it and the results will be presented on Tuesday.

Speakers
Paul Vixie, MFN

Full Abstract

Speakers
Greg Hankins, Riverstone
Susan Harris, Merit Network

Full Abstract

Internet Data centers provide a secure, high-capacity environment for communications companies and content providers to locate and host their Internet equipment. These centers provide the foundation for a variety of business models and Internet operations. The expertise involved in building and operating a network-focused facility are non-trivial. Even the low-end data center requires both an immediate cash draw and a recurring operations cost. The construction project itself requires months of planning and execution. Like most things, the first time one enters into a new arena, potentially costly mistakes are made.

This BOF will explore the tradeoffs between building a data center and outsourcing the data center to a third party. The BOF will also continue some of the discussions that took place during the Data Center BOF at the 1999 Atlanta NANOG.

One goal of this session is to create the framework to establish a list of "public" data centers, listing the services offered, the center's location, and any additional benefits the center offers its customers.

Speakers
Michael Lucking, Epoch Internet
Over the past decade, Michael Lucking has worked with several companies building various-sized data centers. His work has included constructing POPs for regional and national ISPs, and large data centers for government organizations. Michael is currently working as a Sr. Network Engineer for Epoch Internet, Inc.

Bill Norton, Equinix
As Co-Founder and Director of Business Development at Equinix, Bill Norton focuses his attention on building strategic relationships among companies participating at the Internet Business Exchanges. Previously, he was the Chair of NANOG and Manager of the Internet Engineering Group at Merit, leading a variety of national and international network research and operations projects.

Full Abstract

The technology and the economics behind some next-generation commercial IP networks are examined and an implementation of a backbone built on those principles is presented.

Speakers
Jimmy Boyle, Level 3
Jim Boyle works for Level3 Communications, where he has worked on the teams that developed its ATM, and then MPLS, multiservice backbone networks. He is currently seeing if he can load IP software onto DACs and ADMs.

Vijay Gill, MFN
Positions held by Vijay Gill include systems development and analysis at University of Maryland, Baltimore County, and Senior Network Architect/Senior Member of the Technical Staff at UUNET. He is currently Senior Architect at Metromedia Fiber Networks.

Full Abstract

Active queue management (AQM) refers to a family of packet-dropping mechanisms for FIFO queues that has been proposed to support end-to-end congestion control mechanisms in the Internet. We examine the performance of AQM mechanisms by varying two parameters: the queue size and the dropping function. AQM flavors considered include both the well known "Drop from Tail" (TD) and "RED," as well as the more recently proposed "Gentle RED" and a previously unrecognized mechanism we call "Gentle RED with instantaneous queue size", or GRED-I.

We use experimentation to analyze the four AQM mechanisms identified above. The major result observed is that from a TCP goodput and UDP loss standpoint, varying AQM parameters has a minor influence on performance. On the other hand, we observe an important sensitivity of AQM to traffic characteristics. We show that RED and GRED do not provide sufficient performance improvement compared to Tail Drop. In the case of consecutive losses, we show that RED can be harmful by dropping much longer bursts than Tail Drop.

We conclude the presentation by discussing the feasibility of Fair Queueing as an alternative to Active Queue Management. Measurement data collected on the Sprint operational IP backbone will be used to advocate the deployement of Fair Queueing.

Speakers
Christophe Diot, Sprint
Christophe Diot works with the Sprint Advanced Technology Laboratory, where he is in charge of the IP research group. Sprint ATL's charter is to analyze new mechanisms and protocols for the Internet. Current activities in the IP area address diffserv, RED, and PIM-SSM deployment. The IP group's major project is the passive monitoring of the Sprint IP backbone in order to study IP traffic characteristics.

Full Abstract

Link-flooding, as perpetrated by distributed denial of service attacks, is a serious threat to the Internet. We propose a pushback mechanism to defeat such attacks: routers experiencing sustained, massive congestion ask the upstream routers to drop the packets instead, thus freeing bandwidth for other traffic. This algorithm is applied recursively, pushing back to either an uncongested link or even the sources of the problem. Simulation results show that this scheme is indeed effective.

Speakers
Steve Bellovin, AT&T
Steven Bellovin, the co-author of Firewalls and Internet Security: Repelling the Wily Hacker, does research at AT&T Labs on networks, security, and especially why the two don't get along.

Full Abstract

(1) Over a dozen large carriers participate in activities to protect their networks from damage by backhoes and other hazards. What is the latest news from the backhoe damage prevention industry?

(2) An analysis of network outages and events in the Year 2000 compared to previous years shows some changes in the types of outages, and their impact on network users. Are users being impacted by fewer events, or has the number of users grown, diluting the impact of network problems? With little hard data from providers, can we use other measurements of Internet performance to guage the effects?

Speakers
Sean Donelan, Equinix

Full Abstract

This panel discussion will begin with a brief discussion of network infrastructure developments in Columbia and other South American countries. Presentations will then be given by representatives of the two South Florida NAP initiatives: NAP Of the Americas and the Next Generation Internet Exchange, nicknamed the Florida MIX. The focus will be their architectural details, scopes, and positioning in the marketplace. A question-&-answer session follows the individual presentations.

Speakers
Moderator - Bill Manning, ISI.
Panelist - Dan Golding, Netrail
Panelist - Christian Kuhtz, BellSouth

Full Abstract

Some time ago the RIPE NCC embarked in a project to completely rewrite the database software. The project is almost finished and the transition to the new database system will occur in the first half of 2001.

One of the major changes in the new database is its support of RPSL (RFC 2622) and RPS-AUTH (RFC 2725) specifications. The new database also has additional functionality, which requires an extension of the query syntax.

This presentation will discuss RPSL and RPS-AUTH features of the new software and how they will affect the community, give an update on the state of the RIPE IRR, and finally give an overview of IRR-related RIPE activities and rollout plans.

Speakers
Andrei Robachevsky, RIPE-NCC
Andrei Robachevsky is manager of the database group at the RIPE-NCC, where he is responsible for all aspects of the database re-implementation project and RIPE database operations. He has been working in this position for the last year. Prior to this position he worked with the RIPE-NCC as a database software engineer developing software for the re-implementation project. Before he joined the RIPE-NCC in 1999 he was involved in academic network development in Russia.

Full Abstract

A large fraction of OC-192/STM-64 links are between equipment in a single central office (CO). The Optical Internetworking Forum (OIF) recently adopted two Very Short Reach (VSR) OC-192 interface Implementation Agreements to help reduce the cost of these intra-office OC-192 links. VSR-1 is 12 fibers at 1.25 Gbit/s each, using multi-mode ribbon fiber and reaching up to 300 meters. VSR-2 is 10 Gbit/s serial using single mode fiber (SMF) and reaching up to 600m.

This pair was adopted because the parallel optics of VSR-1 will likely be substantially cheaper for the shortest reaches, where the ribbon fiber can be treated like any other patch cable. The serial solution of VSR-2 uses installed SMF and can reach across the largest COs while still offering some potential cost savings compared to SONET short reach (SR) interfaces. The input from the VSR BoF at NANOG-20 was instrumental in reaching this decision.

VSR-1 reaches up to 300 meters and uses 12 lasers signaling at 1.25 Gbit/s. It leverages the 850 nanometer lasers and multi-mode fiber technology used in Gigabit Ethernet. The lasers can be combined in Vertical Cavity Surface Emitting Laser (VCSEL) arrays, and the fibers are joined in a 12-fiber ribbon with a MPO/MTP connector. 10 fibers carry the data, one fiber carries CRC error detection data, and the 12th fiber carries parity of the 10 data fibers. This enables hitless correction of errors on any single fiber, including the loss of a fiber. Field termination is supported, and the link electronics automatically compensate for either ribbon orientation so there is no such thing as a twisted ribbon or a connector put on upside down.

VSR-2 reaches up to 600m and uses serial 10 Gbit/s serial signaling. It is based on ITU G.691 and uses a 1310 nanometer laser over single mode fiber (SMF). Compared to existing SONET Short Reach (SR) interfaces, it relaxes some optical parameters. It allows increased dispersion and reduced reach VSR-1 and VSR-2 are complementary. VSR-1 uses lower speed signals and will likely be the lowest cost OC-192 link for the shortest reaches, where the ribbon fiber can be treated like any other patch cable. VSR-2 uses installed SMF and can reach across very large COs.

Speakers
Russ Tuck, Pluris
Russ Tuck chairs the OIF's Physical and Link Layer Working Group, which is developing VSR Implementation Agreements. He is Systems Architect at Pluris Terabit Network Systems, where he helps lead the development of scalable and reliable core routers. He previously architected scalable database servers at Pyramid Technology, and scalable supercomputers at MasPar. He earned a Ph.D. in Computer Science from Duke University and carried out his dissertation research at UNC Chapel Hill.

Full Abstract

Speakers
Susan Harris, Merit Network