Mul7cast protocols. IP Mul7cast and IGMP SRM (Scalable Reliable Mul7cast) PGM (Pragma7c General Mul7cast)

Transcription

1 IP ANYCAST and MULTICAST; OVERLAYS and UNDERLAYS 1 IP Anycast Outline today Mul7cast protocols IP Mul7cast and IGMP SRM (Scalable Reliable Mul7cast) PGM (Pragma7c General Mul7cast) Overlay networks Tunnels between host computers Build networks on top of the Internet Provide beker control, flexibility, QoS, isola7on, Underlay tunnels Across routers within AS Build networks below IP route Provide beker control, flexibility, QoS, isola7on, 2 1

2 Limita7ons of DNS- based failover Failover/load balancing via mul7ple A records!;; ANSWER SECTION:!! If server fails, service unavailable for TTL Very low TTL: Extra load on DNS Anyway, browsers cache DNS mappings What if root NS fails? All DNS queries take > 3s? 3 Mo7va7on for IP anycast Failure problem: client has resolved IP address What if IP address can represent many servers? Load- balancing/failover via IP addr, rather than DNS IP anycast is simple reuse of exis7ng protocols Mul7ple instances of a service share same IP address Each instance announces IP address / prefix in BGP / IGP Rou7ng infrastructure directs packets to nearest instance of the service Can use same selec7on criteria as installing routes in the FIB No special capabili7es in servers, clients, or network 4 2

3 IP anycast in ac7on Announce / Router 2 Server Instance A Client Router 1 Router Router 4 Server Instance B Announce /32 IP anycast in ac7on Router 2 Server Instance A Client Router 1 Router Router 4 Server Instance B Rou@ng Table from Router 1: Des@na@on Mask Next- Hop Distance / / /

4 IP anycast in ac7on Router 2 Server Instance A Client Router 1 Router Router 4 Server Instance B DNS lookup for hkp:// produces a single answer: IN A IP anycast in ac7on Router 2 Server Instance A Client Router 1 Router Router 4 Server Instance B Rou@ng Table from Router 1: Des@na@on Mask Next- Hop Distance / / /

5 IP anycast in ac7on Router 2 Server Instance A Client Router 1 Router Router 4 Server Instance B Rou@ng Table from Router 1: Des@na@on Mask Next- Hop Distance / / / IP anycast in ac7on Router 2 Server Instance A Client Router 1 Router Router 4 Server Instance B Rou@ng Table from Router 1: Des@na@on Mask Next- Hop Distance / / /

6 IP anycast in ac7on From client/router perspec7ve, topology could as well be: Router 2 Client Router 1 Server Router Router 4 Rou@ng Table from Router 1: Des@na@on Mask Next- Hop Distance / / / Downsides of IP anycast Many Tier- 1 ISPs ingress filter prefixes > /24 Publish a /24 to get a single anycasted address: Poor u7liza7on Scales poorly with the # anycast groups Each group needs entry in global rou7ng table Not trivial to deploy Obtain an IP prefix and AS number; speak BGP Subject to the limita7ons of IP rou7ng No no7on of load or other applica7on- layer metrics Convergence 7me can be slow (as BGP or IGP convergence) Failover doesn t really work with TCP TCP is stateful; other server instances will just respond with RSTs Anycast may react to network changes, even though server online Root name servers (UDP) are anycasted, likle else 12 6

7 Mul7cast protocols 13 Mul7cas7ng messages Simple applica7on mul7cast: Iterated unicast Client simply unicasts message to every recipient Pros: simple to implement, no network modifica7ons Cons: O(n) work on sender, network Advanced overlay mul7cast ( peer- to- peer ) Build receiver- driven tree Pros: Scalable, no network modifica7ons Cons: O(log n) work on sender, network; complex to implement IP mul7cast Embed receiver- driven tree in network layer Pros: O(1) work on client, O(# receivers) on network Cons: requires network modifica7ons; scalability concerns? 14 7

8 IP Mul7cast Simple to use in applica7ons Mul7cast group defined by IP mul7cast address IP mul7cast addresses look similar to IP unicast addrs to (RPC 3171) 265 M mul7cast groups at most Best effort delivery only Sender issues single datagram to IP mul7cast address Routers delivery packets to all subnetworks that have a receiver belonging to the group Receiver- driven membership Receivers join groups by informing upstream routers Internet Group Management Protocol (v3: RFC 3376) 15 IGMP v1 Two types of IGMP msgs (both have IP TTL of 1) Host membership query: Routers query local networks to discover which groups have members Host membership report: Hosts report each group (e.g., mul7cast addr) to which belong, by broadcast on net interface from which query was received Routers maintain group membership Host sends an IGMP report to join a group Mul7cast routers periodically issue host membership query to determine liveness of group members Note: No explicit leave message from clients 16 8

9 IGMP IGMP v2 added: If mul7ple routers, one with lowest IP elected querier Explicit leave messages for faster pruning Group- specific query messages IGMP v3 added: Source filtering: Join specifies mul7cast only from or all but from specific source addresses 17 IGMP Parameters Maximum report delay: 10 sec Query internal default: 125 sec Time- out interval: 270 sec 2 * (query interval + max delay) Ques7ons Is a router tracking each akached peer? Should clients respond immediately to membership queries? What if local networks are layer- two switched? 18 9

10 So far, we ve seen best- effort IP mul7cast 19 Challenges for reliable mul7cast Ack- implosion if all des7na7ons ack at once Source does not know # of des7na7ons How to retransmit? To all? One bad link effects en7re group Only where losses? Loss near sender makes retransmission as inefficient as replicated unicast Once size fits all? Heterogeneity: receivers, links, group sizes Not all mul7cast applica7ons need reliability of the type provided by TCP. Some can tolerate reordering, delay, etc

11 Scalable Reliable Mul7cast Receives all packets or unrecoverable data loss Data packets sent via IP mul7cast ODATA includes sequence numbers Upon packet failure: Receiver mul7casts a NAK or sends NAK to sender, who mul7casts a NAK confirma7on (NCF) Scale through NAK suppression if received a NAK or NCF, don t NAK yourself What do we need to do to get adequate suppression? Add random delays before NAK ing But what if the mul7cast group grows big? Repair through packet retransmission (RDATA) From ini7al sender From designated local repairer (DLR IETF loves acronyms!) 21 Pragma7c General Mul7cast (RFC 3208) Similar approach as SRM: IP mul7cast + NAKs but more techniques for scalability Hierarchy of PGM- aware network elements NAK suppression: Similar to SRM NAK elimina7on: Send at most one NAK upstream Or completely handle with local repair! Constrained forwarding: Repair data can be suppressed downstream if no NAK seen on that port Forward- error correc7on: Reduce need to NAK Works when only sender is mul7cast- able 22 11

12 Overlay Networks 23 Overlay Networks 24 12

13 Overlay Networks Focus at the application level 25 IP Tunneling to Build Overlay Links IP tunnel is a virtual point- to- point link Illusion of a direct link between two separated nodes Logical view: A B E F tunnel Physical view: A B E F Encapsula7on of the packet inside an IP datagram Node B sends a packet to node E containing another packet as the payload 26 13

14 Tunnels Between End Hosts B Src: A Dest: B A Src: C Dest: B Src: A Dest: B Src: A Dest: C C Src: A Dest: B 27 Overlay Networks A logical network built on top of a physical network Overlay links are tunnels through the underlying network Many logical networks may coexist at once Over the same underlying network And providing its own par7cular service Nodes are oven end hosts Ac7ng as intermediate nodes that forward traffic Providing a service, such as access to files Who controls the nodes providing service? The party providing the service Distributed collec7on of end users 28 14

15 Overlays for Incremental Deployment 29 Using Overlays to Evolve the Internet Internet needs to evolve IPv6 Security Mobility Mul7cast But, global change is hard Coordina7on with many ASes Flag day to deploy and enable the technology Instead, beker to incrementally deploy And find ways to bridge deployment gaps 30 15

16 6Bone: Deploying IPv6 over IP4 Logical view: Physical view: A B E F tunnel IPv6 IPv6 IPv6 IPv6 A B C D E F IPv6 IPv6 IPv4 IPv4 IPv6 IPv6 Flow: X Src: A Dest: F data Src:B Dest: E Flow: X Src: A Dest: F Src:B Dest: E Flow: X Src: A Dest: F Flow: X Src: A Dest: F data data data A-to-B: IPv6 B-to-C: IPv6 inside IPv4 B-to-C: IPv6 inside IPv4 E-to-F: IPv6 31 Secure Communica7on Over Insecure Links Encrypt packets at entry and decrypt at exit Eavesdropper cannot snoop the data or determine the real source and des7na7on 32 16

17 Communica7ng With Mobile Users A mobile user changes loca7ons frequently So, the IP address of the machine changes oven The user wants applica7ons to con7nue running So, the change in IP address needs to be hidden Solu7on: fixed gateway forwards packets Gateway has a fixed IP address and keeps track of the mobile s address changes gateway 33 MBone: Mul7cast Backbone A catch- 22 for deploying mul7cast Router vendors wouldn t support IP mul7cast since they weren t sure anyone would use it And, since it didn t exist, nobody was using it Idea: sovware implemen7ng mul7cast protocols And unicast tunnels to traverse non- par7cipants 34 17

18 We saw tunneling on top of IP. What about tunneling below IP? Introducing Mul7- Protocol Label Switching (MPLS) 35 MPLS Overview Main idea: Virtual circuit Packets forwarded based only on circuit iden7fier Source 1 Destination Source 2 Router can forward traffic to the same destination on different interfaces/paths

19 MPLS Overview Main idea: Virtual circuit Packets forwarded based only on circuit iden7fier Source 1 Destination Source 2 Router can forward traffic to the same destination on different interfaces/paths. 37 Circuit Abstrac7on: Label Swapping A 1 2 D Tag Out New A 2 D 3 Label- switched paths (LSPs): Paths are named by the label at the path s entry point At each hop, MPLS routers: Use label to determine outgoing interface, new label Thus, push/pop/swap MPLS headers that encapsulate IP Label distribu@on protocol: responsible for dissemina7ng signalling informa7on 38 19

20 Reconsider security problem 39 Layer 3 Virtual Private Networks Private communica7ons over a public network A set of sites that are allowed to communicate with each other Defined by a set of administra7ve policies Determine both connec7vity and QoS among sites Established by VPN customers One way to implement: BGP/MPLS VPN (RFC 2547) 20

21 Layer 3 BGP/MPLS VPNs VPN A/Site 2 VPN B/Site /16 CE 1 B1 10.2/16 CE A2 10.2/16 P 1 PE 2 CE B2 VPN B/Site 2 CE 2 B1 P 2 BGP to exchange routes CE A1 10.1/16 VPN A/Site 1 PE 1 P 3 CE B3 10.4/16 PE 3 VPN B/Site 3 VPN A/Site 3 Isola@on: Mul7ple logical networks over a single, shared physical infrastructure Tunneling: Keeping routes out of the core CE A3 10.3/16 MPLS to forward traffic 41 High- Level Overview of Opera7on IP packets arrive at provider edge router (PE) PE 2 Des7na7on IP looked up in forwarding table Mul7ple virtual forwarding tables PE 1 PE 3 Datagram sent to customer s network using tunneling (i.e., an MPLS label- switched path) 42 21

22 Virtual Rou7ng and Forwarding Separate tables per customer at each router RFC 2547: Route Dis7nguishers Customer 1 Customer /24 RD: Purple /24 Customer 2 Customer /24 RD: Blue /24 43 Forwarding in BGP/MPLS VPNs Step 1: Packet arrives at incoming interface Site VRF determines BGP next- hop and Label #2 Label 2 IP Datagram Step 2: BGP next- hop lookup, add corresponding LSP (also at site VRF) Label 1 Label 2 IP Datagram 44 22

23 Forwarding PE and P routers have BGP next- hop reachability through the backbone IGP Labels are distributed through LDP (hop- by- hop) corresponding to BGP Next- Hops Two- Label Stack is used for packet forwarding Top label indicates Next- Hop (interior label) Second label indicates outgoing interface / VRF (exterior label) Corresponds to LSP of BGP next-hop (PE) Corresponds to VRF/ interface at exit Layer 2 Header Label 1 Label 2 IP Datagram 45 Forwarding VPN A/Site 2 VPN B/Site 1 CE 1 B1 10.1/16 P /16 CE A2 PE 2 CE B2 10.2/16 VPN B/Site 2 CE 2 B1 P 2 CE A1 10.1/16 VPN A/Site 1 PE 1 PE 3 CE P A /16 CE B3 VPN A/Site /16 VPN B/Site 3 Layer 2 Header Label 1 Label 2 IP Datagram 46 23

24 IP Anycast Outline today Mul7cast protocols IP Mul7cast and IGMP SRM (Scalable Reliable Mul7cast) PGM (Pragma7c General Mul7cast) Overlay networks Tunnels between host computers Build networks on top of the Internet Provide beker control, flexibility, QoS, isola7on, Underlay tunnels Across routers within AS Build networks below IP route Provide beker control, flexibility, QoS, isola7on, 47 24