GENI Research Plan Requirement for GENI

1
GENI Research PlanRequirement for GENI

• 2007.09.21
• Yeongjae Yu
• yuyeongjae_at_gmail.com

2
Table of Contents

• 1. Requirement for GENI
• 1.1 Multiple Simultaneous Experiments
• 1.2 Generality
• 1.3 Support for Real Applications
• 1.4 Support for Real Users
• 1.5 Fidelity
• 1.6 Support for All Aspects of a New Network
  Architecture
• 1.7 Support for Experimenters
• 1.8 Federation Sustainability
• 1.9 Striking a Balance
• 2. Related Work
• 2.1 PlanetLab Virtual Network Infrastructure
• 2.2 User Controlled LightPath Articulated
  Private Network
• 2.3 Relation among PlanetLab, VINI and UCLP

3

• 1. Requirement for GENI

4
1.1 Multiple Simultaneous Experiments

• The concept behind GENI
• - It can be used to multiple ideas and concepts
• To support multiple experiments simultaneously
• gt The concept of slices
• - The resources of GENI can be divided up among
  many different researchers in
• such a way that each can run his own
  experiment
• Virtualization
• - One approach to slices
• - A processor, a fiber link, wireless system,
  ...
• Controlled Isolation
• - GENI must provide strong containment for
  experiments
• - GENI must support controlled interconnection
  of slices to each other and to the
• current Internet

5
1.2 Generality

• We want to build an experimental platform that
  can support
• a wide range of future Internets
• How much generality is required to support the
  anticipated
• experiments?
• - To experiment with packet formats that
  materially differ from those of the
• Internet
• - To move beyond the paradigm of packet
  switching and explore other modes for
• sharing and resource allocation
• - To exploit specific features of the different
  technologies included in GENI
• - To experiment with architectures that include
  network-level operations other
• than simple packet forwarding
• Diversity of technology
• Balancing generality with cost

6
1.3 Support for Real Applications

• GENI should be able to support not just a future
  network,
• but also the applications that might run on
  that network
• To attract real applications, the GENI must
  include facilities
• for development and deployment of applications,
  not just
• data transport

7
1.4 Support for Real Users

• If we are to gain real experience with real
  applications, we
• must allow real users to try them out, and make
  real use of
• them
• What does it mean to support real users?
• - The GENI facility must reach to the edge of
  the network, where the users
• connect
• - There must be a rich connectivity between
  GENI and the Internet of today
• - There must be an adequate pool of potential
  users that have end-node
• computers directly connected to, and a part
  of, the GENI infrastructure
• - Some support for slices needs to be provided
  for the end-nodes that are
• attached to GENI

8
1.5 Fidelity

• Reach
• - As wide a reach as possible
• Topology
• - Keeping delays within a small factor of
  physical distance
• - Path diversity
• - Underlying fiber paths
• - Major interconnection points (exchange
  points, aggregation points)
• Realism of virtualization
• Physical distribution
• Scale
• Failure modes
• - Intentionally induced failure, unanticipated
  failure

9
1.6 Support for All Aspects of a New Network
Architecture

• Support for management
• - It must important that the management aspects
  of all devices be fully virtualized
• - We can have the equivalent of virtual system
  operators
• Support for security
• - The GENI infrastructure itself must be stable
  and secure
• - The mechanisms for isolation among slices
  must be very robust
• - There may be a requirement for specialized
  security technology
• Support for anticipated future capabilities
• - In 10 years, there may be features that will
  be commonplace then, but are not
• yet realized in any effective way

10
1.7 Support for Experimenters

• Ease of Use
• - GENI must remove as many practical barriers
  as possible to researchers being
• able to make full use of the facility
• Observability
• - GENI must offer strong support for
  measurement-based quantitative research
• Fail-safe
• - GENI must be secure, so that its resources
  cannot accidentally or maliciously
• be used to attack todays Internet
• Sources of real traffic
• - GENI must provide a way experiments can be
  run with real traffic
• - One approach to have several large-scale
  popular services
• ex) content distribution networks

11
1.8 Federation Sustainability

• GENI must be designed for a 15-20 year lifetime
• To ensure the sustainability
• - Support for federation
• - Design with operational costs in mind
• Addition of new technology
• - Open hardware interfaces
• - The ability to virtualize devices
• - The ability to incorporate new devices into
  the GENI management mechanisms
• Living in the future
• - GENI is supposed to be a tolerably realistic
  emulation of a networking
• technology base 10 years in the future

12
1.9 Striking a Balance

• What makes GENI a unique and compelling
  instrument is
• how it balances requirements to support
  research that
• simply cannot be done today
• - Resolving conflicts among requirements (ex.
  Sliceability vs Fidelity)
• - Recognizing the specific combination of
  capabilities that are unique to GENI
• (1) wide-spread deployment
• (2) a diverse and extensible collection of
  network technologies
• (3) support for real user traffic

13

• 2. Relate Work

14
2.1 PlanetLab VINI

• PlanetLab
• - PlanetLab is a global overlay network for
  developing and accessing broad-
• coverage network services Chun 03
• - PlanetLab allows multiple services to run
  concurrently and continuously, each in
• its own slice of PlanetLab Chun 03
• Slice A horizontal cut of global resources
  Chun 03
• The substrate resources bound to
  a particular experiment Clark 07
• Virtual Network Infrastructure (VINI)
• - VINI is a virtual network infrastructure that
  allows network researchers to
• evaluate their protocols and services in a
  realistic environment that also
• provides a high degree of control over
  network conditions.
• - PL-VINI is a prototype of a VINI that runs on
  the public PlanetLab. PL-VINI
• enables arbitrary virtual networks,
  consisting of software routers connected by
• tunnels, to be configured within a PlanetLab
  slice.

15
2.2 UCLP APN

• User Controlled LightPath (UCLPv2)
• - UCLP is a network virtualization management
  tool built using web services
• Lemay 06
• ex) XC-WS(Cross Connect Web Service) for
  SONET, SDH and Lambda Cross
• Connects
• - Users can create several parallel
  application specific networks from a single
• physical network through UCLP Lemay 06
• Articulated Private Network (APN)
• - An aggregate mix of resources St.Arnaud 07

16
2.3 Relation among PlanetLab, VINI and UCLP
ltCurrent Relationgt
ltTarget Relationgt