3-1 Design Principles Goals: r identify, study common architectural components, protocol mechanisms r what approaches do we find in network architectures? r synthesis: big picture design principles: r separation of data, control r hard state versus soft state ü randomization ü indirection r network virtualization: overlays r multiplexing r design for scale

3-2 Virtualization of networks Virtualization of resources: powerful abstraction in systems engineering: r computing examples: virtual memory, virtual devices m virtual machines: e.g., java m IBM VM os from 1960s/70s r layering of abstractions: dont sweat the details of the lower layer, only deal with lower layers abstractly

3-3 Examples r Connecting local heterogeneous networks r IP over ATM r Resilient overlay networks r VPN

3-4 The Internet: virturalizing local networks 1974: multiple unconnected networks m ARPAnet m data-over-cable networks m packet satellite network (Aloha) m packet radio network.. differing in: m addressing conventions m packet formats m error recovery m routing

3-5 Cerf & Kahn: Interconnecting two networks r …interconnection must preserve intact the internal operation of each network. r..the interface between networks must play a central role in the development of any network interconnection strategy. We give a special name to this interface that performs these functions and call it a GATEWAY. r.. prefer that the interface be as simple and reliable as possible, and deal primarily with passing data between networks that use different packet-switching strategies r …address formats is a problem between networks because the local network addresses of TCP's may vary substantially in format and size. A uniform internetwork TCP address space, understood by each GATEWAY and TCP, is essential to routing and delivery of internetwork packets. ARPAnet satellite net

3-6 Cerf & Kahn: Interconnecting two networks ARPAnet satellite net Gateway: r embed internetwork packets in local packet format or extract them r route (at internetwork level) to next gateway gateway Internetwork layer: r addressing: internetwork appears as a single, uniform entity, despite underlying local network heterogeneity r network of networks

3-7 Historical Aside: Proposed Internetwork packet in 1974: local header source address dest. address seq. # byte count flag field text checksum network TCP identifier 8 16

3-8 Cerf & Kahns Internetwork Architecture What is virtualized? r two layers of addressing: internetwork and local network r new layer makes everything homogeneous at internetwork layer r underlying local network technology (cable, satellite, 56K modem) is invisible at internetwork layer

3-9 IP-Over-ATM Classic IP only r 3 networks (e.g., LAN segments) r MAC (802.3) and IP addresses IP over ATM r replace network (e.g., LAN segment) with ATM network r ATM addresses, IP addresses ATM network Ethernet LANs Ethernet LANs

3-10 IP-Over-ATM AAL ATM phy Eth IP ATM phy ATM phy app transport IP AAL ATM phy app transport IP Eth phy

3-11 IP View of the world ATM network IP network

3-12 Classical IP-over ATM [RFC 1577] A B C D E R1 R2 LIS: logical IP subnet r end systems in same LIS have same IP network addr r LIS looks like a LAN r ATM net divided into multiple LIS r Intra-LIS communication via direct ATM connections m How to go from IP addr to ATM addr: ATMARP resolves IP addr to ATM addr (similar to ARP) LIS 1 LIS 2 LIS 3

3-13 Classical IP-over ATM [RFC 1577] A B C D E R1 R2 Inter-LIS communication: r source, dest. in different LIS r each LIS looks like a LAN r hop-by hop forwarding: m A-R1-R2-E LIS 1 LIS 2 LIS 3

3-14 NHRP (next hop resolution protocol) [RFC 2332] r source/dest. not in same LIS: ATMARP can not provide ATM dest. address r NHRP: resolve IP-to-ATM address of remote dest. m client queries local NHRP server m NHRP server routes NHRP request to next NHRP server m destination NHRP returns dest ATM address back through NHRP server chain (like routed DNS) r source can send directly to dest. using provided ATM address A B C D E NHRP server, S 1 LIS 1 LIS 2 LIS 3 NHRP server, S 2 NHRP server, S 3 ARP over multiple hops

3-15 IP-over-ATM: why? r because its there- use ATM network as a link-layer to connect IP routers r can manage traffic more carefully in ATM network (e.g., rate-limit source/dest pairs, provide CBR service) r leave IP untouched – leverage the fact that many users have IP addresses already

3-16 Resilient Overlay Networks Overlay network: r applications, running at various sites as nodes on an application-level network r create logical links (e.g., TCP or UDP connections) pairwise between each other r each logical link: multiple physical links, routing defined by native Internet routing

3-17 Overlay network

3-18 Overlay network Focus at the application level

3-19 Internet Routing r BGP defines routes between stub networks UCLA Noho.net Berkeley.net UMass.net Internet 2 Mediaone C&W

3-20 Internet Routing r BGP defines routes between stub networks UCLA Noho.net Berkeley.net UMass.net Internet 2 Mediaone C&W Noho-to-UMass

3-21 Internet Routing r BGP defines routes between stub networks UCLA Noho.net Berkeley.net UMass.net Internet 2 Mediaone C&W Noho-to-Berkeley

3-22 Internet Routing UCLA Noho.net Berkeley.net UMass.net Internet 2 Mediaone C&W Noho-to-Berkeley Congestion or failure: Noho to Berkely BGP- determined route may not change (or will change slowly)

3-23 Internet Routing UCLA Noho.net Berkeley.net UMass.net Internet 2 Mediaone C&W Noho-to-Berkeley Noho to UMass to Berkeley r route not visible or available via BGP! r MediaOne cant route to Berkeley via Internet2 Congestion or failure: Noho to Berkely BGP- determined route may not change (or will change slowly)

3-24 RON: Resilient Overlay Networks Premise: by building application overlay network, can increase performance, reliability of routing Two-hop (application-level) noho-to-Berkeley route application-layer router Virtualize the Internet! Layer 7 routing!

3-25 RON Experiments r measure loss, latency, and throughput with and without RON r 13 hosts in the US and Europe r 3 days of measurements from data collected in March 2001 r 30-minute average loss rates m A 30 minute outage is very serious! r Note: Experiments done with No- Internet2-for-commercial-use policy

3-26 An order-of-magnitude fewer failures % % % % % % RON Worse No Change RON Better Loss Rate 30-minute average loss rates 6,825 path hours represented here 12 path hours of essentially complete outage 76 path hours of TCP outage RON routed around all of these! One indirection hop provides almost all the benefit! 6,825 path hours represented here 12 path hours of essentially complete outage 76 path hours of TCP outage RON routed around all of these! One indirection hop provides almost all the benefit!

3-27 RON Research Issues how to design overlay networks? Measurement and self-configuration Fast fail-over Sophisticated metrics application-sensitive (e.g., delay versus throughput) path selection effect of RON on underlying network If everyone does RON, are we better off? Interacting levels of control (network- and application-layer routing

3-28 Virtual Private Networks (VPN) r SP infrastructure: m backbone m provider edge devices r Customer: m customer edge devices (communicating over shared backbone) Networks perceived as being private networks by customers using them, but built over shared infrastructure owned by service provider (SP) VPNs

3-29 VPN reference architecture customer edge device provider edge device

3-30 VPN: logical view customer edge device provider edge device virtual private network

3-31 Leased-line VPN customer sites interconnected via static virtual channels (e.g., ATM VCs), leased lines customer site connects to provider edge

3-32 Customer premise VPN customer sites interconnected via tunnels tunnels encrypted typically SP treats VPN packets like all other packets All VPN functions implemented by customer

3-33 Drawbacks r Leased-line VPN: configuration costs, maintainence by SP: long time, much manpower r CPE-based VPN: expertise by customer to acquire, configure, manage VPN Network-based VPN r customers routers connect to SP routers r SP routers maintain separate (independent) IP contexts for each VPN m sites can use private addressing m traffic from one vpn can not be injected into another

3-34 Network-based Layer 3 VPNs multiple virtual routers in single provider edge device

3-35 Tunneling

3-36 VPNs: why? r privacy r security r works well with mobility (looks like you are always at home) r cost: many forms of newer VPNs are cheaper than leased line VPNs m ability to share at lower layers even though logically separate means lower cost m exploit multiple paths, redundancy, fault-recovery in lower layers m Need isolation mechanisms to ensure resources shared appropriately r abstraction and manageability: all machines with addresses that are in are trusted no matter where they are