Network Layer Protocols

Transcription

1 internetwork n. &v. Network Layer Protocols CSCI 363 Computer Networks Department of Computer Science 1 logical network built out of a collection of physical networks. 2 tr. to interconnect physical networks into a single logical networks; to build an internet. CSCI 363 Computer Networks 2 Simple internetwork (example) Service model Question: What are the challenges in connecting two physical networks into a single logical network? Host-to-host service Philosophy: Define a model that is so undemanding that almost any network technology is able to provide the required service. (IP) Addressing scheme Datagram Best-effort Questions: What does best-effort mean? How does it simplify internetworking? CSCI 363 Computer Networks 3 CSCI 363 Computer Networks 4 Data delivery model Data delivery model Every datagram contains enough information for the network to forward it from source to destination. Question: Why is Version at the start of the packet? Aligned on 32-bit boundaries. HLen: length of header. TOS: type of service. Length: length of datagram. Questions: What is the maximum length of a datagram? Will the underlying physical network support it? Aligned on 32-bit boundaries. CSCI 363 Computer Networks 5 CSCI 363 Computer Networks 6

2 TTL: time to live (hop count). Protocol: identity of upperlevel protocol to which IP passes this packet. Checksum: applies to header. SourceAddr: to allow filtering based on source. DestinationAddr: used for data delivery. Options: header length determines if any are set. Data delivery model Aligned on 32-bit boundaries. CSCI 363 Computer Networks 7 Fragmentation and Reassembly Ethernet: max. frame length = 1500 B. MTU: The largest IP datagram that a network can carry in a single frame. Question: How does the underlying network technology affect IP? How does one choose the maximum datagram size for IP? Design options: (1) Choose IP datagram max. length to be compatible with MTU. (2) Fragment IP datagram into multiple MTUs. CSCI 363 Computer Networks 8 MTU Fragmentation and Reassembly Fragmentation and Reassembly Gotcha: MTU header Payload: IP datagram trailer 1. Each fragment is itself a self-contained IP datagram that is transmitted over a sequence of physical networks, independent of other fragments. 2. Each IP datagram is reencapsulated for each physical network over which it travels. CSCI 363 Computer Networks 9 CSCI 363 Computer Networks 10 Fragmentation and Reassembly (a) (b) Ident = x Ident = x Ident = x Ident = x Start of header 0 Rest of header 1400 data bytes Start of header 1 Rest of header 512 data bytes Start of header 1 Rest of header 512 data bytes Start of header 0 Rest of header 376 data bytes Offset = 0 Offset = 0 Offset = 64 Offset = 128 Some nitty-gritty details If a fragment is lost, the receiver still tries to assemble the datagram When this fails, it has to clean up the memory that was used in the attempt. If hosts can do path MTU discovery, then IP fragmentation can be avoided altogether. CSCI 363 Computer Networks 11 Global addresses Question: What are the challenges? Wish List: Addresses are unique. Address structure makes routing/forwarding as easy as possible. CLASS A CLASS B CLASS C CSCI 363 Computer Networks 12

3 Virtual Private Networks Implementing a VPN Reality: Two private corporate networks. Traffic from one network is invisible to the other network (ideal privacy). Leased lines, however, may be too expensive to keep. Virtual: Two corporate networks sharing some of the same physical links. Want to give the functionality and the privacy of private networks for each corporation. Question: What are the challenges to making this happen? Encapsulate traffic from R1 to R2 inside IP packets addressed to R2. Together with encryption, this tunneling of packets is an effective way to implement a VPN. Question: Can you think of other uses for tunneling? Routing table NetworkNum NextHop 1 Interface 0 2 Virtual Interface 0 Default Interface 1 CSCI 363 Computer Networks 13 CSCI 363 Computer Networks 14 Routing vs. Forwarding Definition: forwarding consists of taking a packet, identifying its destination, looking up in a table the interface that leads to the destination, and sending it off in that direction. Definition: routing is the process by which forwarding tables are built. Network as a Graph Routing is a problem of graph theory. You are given a set of nodes V and a set of edges E such that e= (v i,v j ) in E for v i and v j in V. These two sets define a graph G= (V,E). You are also given a weight function w(.), which maps each e in E to some real valued weight. Routing table NetworkNum NextHop vs. Forwading table NetworkNum Iface MAC addr 10 eth0 8:0:2:e4:b:1:2 The problem is to find the lowest-cost path between any two nodes, where the cost of a path equals the sum of the weights of all edges in the path. This can be re-stated as the all-pairs shortest path problem. CSCI 363 Computer Networks 15 CSCI 363 Computer Networks 16 CSCI 311 Flashback: Single-Source Shortest Paths In a single-source shortest paths problem we are given a weighted, directed graph G=(V,E), with a weight function on E that maps edges to real valued weights. The weight of a path is defined as the sum of its constituent edges: Proposal Each node discovers its neighbors and tells one specific node (the leader) what they find. The leader runs an algorithm to solve the all-pairs shortest path problem, computing routing tables for each node. The leader communicates the routing table to each node, which stores it in nonvolatile memory in each of the nodes. A shortest path from vertex u to vertex v is defined as any path p with weight: Question: What are the problems with this method? We will consider this problem only for non-negative weights. CSCI 363 Computer Networks 17 CSCI 363 Computer Networks 18

4 Distance Vector Routing Distance Vector Routing A s view: vector Dest Cost NextHop B 1 B C 1 C D - E 1 E F 1 F G - Global view Info Distance to node at node A B C D E F G A B C D E 1 0 F G CSCI 363 Computer Networks 19 Every node starts by building it s own local view of what nodes are 1 hop away. Next, every node sends its vector to its directly connected neighbors. F tells A that it can reach G at cost 1. A knows it can reach F at cost 1, so it updated its own vector to indicate that it can reach G at cost 2. If A were to discover another route to G at a cost higher than 2, it would ignore it and leave its vector as it is. After a few iterations of these exchanges, the routing table converges to a consistent state. Question: How would this method deal with link failures? CSCI 363 Computer Networks 20 Distance Vector Routing The Count-to-infinity Problem Question: How can you detect that a node has failed? Periodic updates: Every t seconds, send your local info to your neighbors. This allows other nodes to know that you are running. Triggered updates: Every time you learn new information from a neighbor that leads you to update your local vector, you send the recomputed vector to all your neighbors. The routing table doesn t converge or stabilize. Link (A,E) goes down. A periodic update kicks in and A advertises that its distance to E is infinity. At about the same time, B and C tell each other that they can reach E in 2 hops. B knows now that it can t communicate with E via A, but concludes that it can send traffic to C which says it can reach A in 2 hops. Now, B is thinking that its distance to E is 3 hops and it lets A know about that. Argh, now A is going to think it can reach E in 4 hops and it will tell C of this fact Where does this end? CSCI 363 Computer Networks 21 CSCI 363 Computer Networks 22 Solutions for 2-node loops Count-to-infinity: Cap infinity at some maximum number of hops that allow a packet to go all the way across the network. Reduce time to convergence: Split-horizon when a node sends a routing update to its neighbors, it does not send the routes it learned from a neighbor back to that neighbor. Reduce time to convergence: Split-horizon with poison reverse communicate back to the sending neighbor but poison the route with negative information (infinity) so that it doesn t end up used as intermediate node in a route. CSCI 363 Computer Networks 23 Link State Routing Assumptions: Each node can discover the state of the link to its neighbors and the cost of each link. Basic principle: If every node knows how to reach its directly connected neighbors, one can spread the local knowledge of each node to all other nodes in the network so that every node can construct a the network weighted graph. With this knowledge, a node can always determine the shortest path to any other node in the network. CSCI 363 Computer Networks 24

5 Link State Routing Mechanisms Reliable dissemination of link-state information! flooding. Calculation of routes from the sum of all the accumulated link-state knowledge. Reliable Flooding Question: If you are a node and you want to tell your neighbors the state of the links incident to you, what should be contained in the information packets that you pass to them?! Underlying Problems: A packet should reflect the state of your links at a given point in time. You need to ensure that old link-state information eventually stops circulating around the network. CSCI 363 Computer Networks 25 CSCI 363 Computer Networks 26 Reliable Flooding Question: How can one guarantee that the linkstate packets sent from a node will definitely arrive at its neighbors? Question: How does one deal with the possibility of having multiple, contradictory copies of a node s link-state packets circulate the network at the same time? CSCI 363 Computer Networks 27 Link-State Packet (LSP) CreatorID: identity of the packet creator. NeighborsList: a list of tuples like <NeighborId, link cost>. SequenceNumber: a counter value that places this packet in the context of all LSPs sent out by the creator of this packet. TTL: time to live in number of hops. Question: When should a node send out an LSP? CSCI 363 Computer Networks 28 Dealing with LSPs An LSP is sent out when: A periodic time expires. A topology change is detected.! Important: Routing messages, that is LSPs, are overhead in the network and they should be kept to a minimum. CSCI 363 Computer Networks 29 Assumptions: Route Calculation (based on Dijkstra s shortest-path algorithm) A node constructs a graph to represent the network to the best of its knowledge (received LSPs). N: V, number of nodes. l(i,j): Non-negative weight of edge (i,j). M: nodes set of nodes incorporated so far for some node s. C(n): cost of the path from node s to node n. Algorithm for a node s: M = {s} for each n in N-{s} do: C(n)=l(s,n) while (N not equal M) M=M U{w} such that C(w)=min{for all w in (N-M)} for each n in (N-M) do: C(n)=min{C(n), C(w)+l(w,n)} CSCI 363 Computer Networks 30

6 Properties of Link-State Routing Stabilizes quickly. Keeps routing control traffic low. Responds rapidly to topology changes. Doesn t scale well: the amoung of information stored in each node is large. Comparison of routing algorithms: Link state vs. Distance Vector CSCI 363 Computer Networks 31 CSCI 363 Computer Networks 32 Distance Vector Routing Every node starts by building it s own local view of what nodes are 1 hop away. Next, every node sends its vector to its directly connected neighbors. F tells A that it can reach G at cost 1. A knows it can reach F at cost 1, so it updated its own vector to indicate that it can reach G at cost 2. If A were to discover another route to G at a cost higher than 2, it would ignore it and leave its vector as it is. After a few iterations of these exchanges, the routing table converges to a consistent state. Link State Routing Assumptions: Each node can discover the state of the link to its neighbors and the cost of each link. Basic principle: If every node knows how to reach its directly connected neighbors, one can spread the local knowledge of each node to all other nodes in the network so that every node can construct a the network weighted graph. With this knowledge, a node can always determine the shortest path to any other node in the network. CSCI 363 Computer Networks 33 CSCI 363 Computer Networks 34 Properties of Link-State Routing Stabilizes quickly. Keeps routing control traffic low. Responds rapidly to topology changes. Doesn t scale well: the amount of information stored in each node is large. Subnetting Question: What makes it stabilize quickly? Question: What can be done to improve scalability? Question: What does it take to mess up routing for everyone? CSCI 363 Computer Networks 35 CSCI 363 Computer Networks 36

7 Global addresses Scenario: Organization X builds local area network with 27 hosts. After 29 days of operation in isolation from the global Internet, Organization X decides that it needs to be on the Internet. Question: How can Organization X join the Internet? CLASS A CLASS B CLASS C Dealing with a network Question: So, Organization X received a network identifier What class of network should it be given to minimize the waste of global IP addresses? What if it has 257 hosts? CLASS A CLASS B CLASS C Question: How would Network Address Translation help? CSCI 363 Computer Networks 37 CSCI 363 Computer Networks 38 Network Address Translation (NAT) local addresses global addresses Subnetting Take a network address and break it up into subnets that can be assigned to individual physical networks globally unique NAT Box Question: What does NAT do to IPs service model? Question: What happens to application messages that contain IP addresses? CSCI 363 Computer Networks 39 Define a subnet mask to help create a new level of hierarchy in the addressing scheme. The bitwise AND of the subnet mask with the full address gives the subnet number. Example: Take host address and subnet mask and compute the subnet number. CSCI 363 Computer Networks 40 Example of Subnetting All hosts in a subnet are configured with the same subnet mask. Question: What does subnetting do to routing as we know it? What happens to the format of forwarding tables? Question: What does subnetting do to ARP? SubnetNumber SubnetMask NextHop eth eth R2 CSCI 363 Computer Networks 41 Datagram Forwarding Algorithm D = destination IP address for each forwarding table entry <SubnetNumber,SubnetMask,NextHop> D1 = SubnetMask & D if (NextHop is an interface) deliver datagram directly to destination else deliver datagram to NextHop (a router) CSCI 363 Computer Networks 42

8 Fine Points on Subnetting The subnet mask does not need to align on byte boundaries. (You don t even have to have contiguous 1 s, although that is not recommended.) It is possible to put multiple subnets on the same physical network, but hosts on the same physical network may then have to go through a router to talk to each other. From outside the subnetted domain, the whole thing is viewed as a single network. For this reason, subnets should be kept geographically close. CSCI 363 Computer Networks 43 Global addresses Scenario: Organization X builds local area network with 27 hosts. After 29 days of operation in isolation from the global Internet, Organization X decides that it needs to be on the Internet. Question: How can Organization X join the Internet? CLASS A CLASS B CLASS C CSCI 363 Computer Networks 44 Dealing with a network Question: So, Organization X received a network identifier What class of network should it be given to minimize the waste of global IP addresses? What if it has 257 hosts? CLASS A Network Address Translation (NAT) local addresses global addresses globally unique NAT Box CLASS B CLASS C Question: How would Network Address Translation help? CSCI 363 Computer Networks Question: What does NAT do to IPs service model? Question: What happens to application messages that contain IP addresses? CSCI 363 Computer Networks 46 Address Assignment Efficiency Organization X requests IP addresses for its physical network with 10 hosts. It receives a class C address. Question: What is the address assignment efficiency? CLASS A CLASS B CLASS C What if you could create a virtual network to contain only as many addresses as the hosts in Organization X? Address Assignment Efficiency 2-host network: assigned a Class C address! Efficiency: 2/254 = 0.78% 256-host network: assigned a Class B address! Efficiency: 256/65,534 = 0.039% How efficiently is the IP address space being used? CSCI 363 Computer Networks 47 CSCI 363 Computer Networks 48

9 Subnetting Example of Subnetting Take a network address and break it up into subnets that can be assigned to individual physical networks. Define a subnet mask to help create a new level of hierarchy in the addressing scheme. The bitwise AND of the subnet mask with the full address gives the subnet number. Example: Take host address and subnet mask and compute the subnet number. eth0 eth1 All hosts in a subnet are configured with the same subnet mask. Question: What does subnetting do to routing as we know it? What happens to the format of forwarding tables? R1 SubnetNumber SubnetMask NextHop eth eth R2 CSCI 363 Computer Networks 49 CSCI 363 Computer Networks 50 Datagram Forwarding Algorithm D = destination IP address for each forwarding table entry <SubnetNumber,SubnetMask,NextHop>! D1 = SubnetMask & D! if (NextHop is an interface)!! deliver datagram directly to destination! else!! deliver datagram to NextHop (a router) Fine Points on Subnetting The subnet mask does not need to align on byte boundaries. (You don t even have to have contiguous 1 s, although that is not recommended.) It is possible to put multiple subnets on the same physical network, but hosts on the same physical network may then have to go through a router to talk to each other. From outside the subnetted domain, the whole thing is viewed as a single network. For this reason, subnets should be kept geographically close. CSCI 363 Computer Networks 51 CSCI 363 Computer Networks 52 Autonomous Systems and Interdomain Routing Classless Routing In order to provide network scalability, we approach the problem of routing using a hierarchical view of the system. We define domains or autonomous systems as interconnected collections of physical networks. Inside an AS, routers use protocols like RIP or OSPF. Once we interconnect ASs, we need to define another level of routing. CSCI 363 Computer Networks 53 CSCI 363 Computer Networks 54

10 Possible Policies on Class B Address Assignment Scenario: An Autonomous System (AS) requests a Class B address. 1) If number of hosts is close to 64K, grant request, otherwise deny it. 2) Instead of giving one Class B address, give multiple Class C addresses to cover all the hosts. CSCI 363 Computer Networks 55 The Multiple Class C Addresses Solution Scenario: One AS called X! 16 Class C network numbers. Question: How many entries for AS X will appear in an Internet backbone router? Question: What if we had given it only one Class B network? What would the address assignment efficiency be? Question: How could we reduce the impact of multiple Class C addresses on routing table size? CSCI 363 Computer Networks 56 Classless Interdomain Routing (CIDR) Goal: Minimize the number of routes that a router needs to know while keeping address assignment efficiency high by giving out contiguous blocks of Class C addresses. Example: top 20 bits remain constant CLASS B: 16 bits Host 20 bits 12 bits Network CLASS C: 8 bits Host Host CIDR In order to make this work, we need to: Assign contiguous blocks of Class C addresses. Each block must contain a number of class C networks that is a power of 2. Use interdomain routing that understands classless addresses (network number can be of any length in number of bits). Routing protocols see network numbers as: <length, value> This is analogous to subnetting: <mask, value> CSCI 363 Computer Networks 57 CSCI 363 Computer Networks 58 CIDR vs. Subnetting Enhanced Route Aggregation Multiple network numbers Multiple physical networks Physical Network One address Physical Network 1 Physical Network One network number Route aggregation, or supernetting. Physical Network 3 Both X and Y are accessible via the same provider network, the ISP advertises a single route for both using the 19-bit prefix they share. If addresses are assigned carefully, better route aggregation becomes possible. CSCI 363 Computer Networks 59 CSCI 363 Computer Networks 60

11 Interdomain Routing: The Border Gateway Protocol domain domain Autonomous Systems Goal: To hierarchically aggregate routing information in a large internetwork to improve scalability. Border routers: this is the default destination of all packets bound to hosts outside the AS. The AS model decouples the intradomain routing that happens within one AS from what is done in another AS. Prefix: as in classless routing, this is expressed as IP/length, i.e /20. CSCI 363 Computer Networks 61 CSCI 363 Computer Networks 62 The BGP Model of the Internet BGP Routing Peering point Consumer ISP Small corporation Large corporation Backbone service provider Large corporation Stub AS: Only one connection to another AS. Consumer ISP Consumer ISP Peering point Multihomed AS: An AS with connections to multiple ASs, which refuses to carry external traffic. Transit AS: An AS with connections to multiple Ass, which carries internal and external traffic. CSCI 363 Computer Networks 63 Challenges: Scale (over 140,000 prefixes), Domains are autonomous and can assign arbitrary metrics to its internal paths, ASs can only cooperate if they can trust one another to publicize accurate routing information and to carry out their promises, Policies should be flexible to allow ASs freedom of action. This choice may override optimal paths and determine the use of paths that are good enough. CSCI 363 Computer Networks 64 AS1 TCP AS2 TCP TCP BGP Routing AS3 Each domain determines at least one node to be it s BGP speaker (border gateways need not be the same nodes). BGP advertises complete paths (enumerated lists of ASs) to reach a certain network. This enables policies to be tailored to the AS wishes and also makes it easy to detect routing loops. AS identifications are unique 16- bit values assigned by a central authority. An AS advertises routes that are good for itself: the speaker picks a favorite. A speaker is under no obligation to advertise routes that it doesn t want used. Speakers can withdraw routes. Integrating Interdomain and Intradomain Routing Stub AS: The border router injects a default route into the intradomain routing protocol. Multihomed and Transit ASs: The border routers inject routes that they have learned from outside the AS. Transit ASs: The information learned from BGP may be too much to inject into the intradomain protocol: if a large number of prefixes in inserted, large link-state packets will be circulated and path calculations will get very complex. CSCI 363 Computer Networks 65 CSCI 363 Computer Networks 66

12 Addresses and Routing 128-bit addresses (IPv4: 32-bit addresses). IPv6 1,500 IPv6 addresses per square foot for the whole planet! CSCI 363 Computer Networks 67 CSCI 363 Computer Networks 68 Address Space Allocation Prefix Use Reserved Unassigned Reserved for NSAP allocation Reserved for IPX allocation , , 0001 Unassigned 1 Aggregatable Global Unicast Classes A, B, and C 010, 011, 100, 101, 110, 1110, Unassigned , , , Link local use May not be unique Site local use May not be unique Multicast Class D CSCI 363 Computer Networks 69 Address Notation X:X:X:X:X:X:X:X where X is a 16-bit value When there are many consecutive 0s, omit them: 47CD:0000:0000:0000:0000:0000:0000:A456:0124 becomes 47CD::A456:0124 (double colon means a group of 0s) Two types of IPv6 address can contain embedded IPv4 addresses. For example, an IPv4 host address becomes ::FFFF: (the last 32 bits are an IPv4 address) This notation facilitates the extraction of an IPv4 address from an IPv6 address. CSCI 363 Computer Networks 70 Aggregatable Global Unicast Addresses 001 prefix: How are these addresses assigned to ISPs, autonomous systems, networks, hosts, and routers? Subscriber: non-transit AS (stub and multihomed) Provider: transit AS Direct: connected directly to subscribers Indirect or Backbone network: connected to other providers Plan: Aggregate routing information to reduce the burden on intradomain routers. Assign a prefix to a direct provider; within the provider assign longer prefixes that reach its subscribers. Aggregatable Global Unicast Addresses Plan: Aggregate routing information to reduce the burden on intradomain routers. Assign a prefix to a direct provider; within the provider assign longer prefixes that reach its subscribers. Question: What is the drawback in this plan? Question: How can this drawback be addressed? CSCI 363 Computer Networks 71 CSCI 363 Computer Networks 72

13 Packet Format IPv4 IPv6 CSCI 363 Computer Networks 73