Network Architecture R 02 L 1 Jon Crowcroft
- Slides: 20
Network Architecture (R 02) - L 1 Jon Crowcroft, http: //www. cl. cam. ac. uk/~jac 22 http: //www. cl. cam. ac. uk/teaching/1213/R 02/
Course Structure 16 Lectures several guest slots - n n n n Yury Audzivich - router h/w algorithmics Christos Efstrathiou - sensor/mobile net arch Dirk Trossen - pub/sub in the net Hamed Haddadi - topology over time Sue Moon – Packet Shader Participation by you n n n Reading & Critique-ing papers See “How to read a paper” by Keshav http: //www. sigcomm. org/ccr/drupal/files/p 83 -keshav. A. pdf
Course Assessment n Your involvement each week n n 3 essays - compare/contrast n n n Alternate me present & you all contribute Due dates Nov 2, Nov 30, start of next term (Jan 18) An annotated bibligraphy n At start of next term, but update each week
Workload n Read 1 -2 papers per week n n Plus scan related material Keep notes Feel free to ask me more Essays can be n n n 2 -4 pages Note form is fine References/citing source material essential
Review of Internet Architecture • Packet switching • • Datagram Network • • (video download can be faster than viewing) “Stateless” • • No set up - fast for transactions Work Conserving • • No circuit, virtual or otherwise (end and router don’t share state) (max pkt size unchanged for 30 yrs!)
Parsimony n End to end model n n (clark et al) Cautious Sender n n Forgiving Receiver (postel principle) Many different kinds of applications and higher-level protocols IP Many different kinds of networks The “Hourglass Model”, Steve Deering
IP packet Vers H. Len TOS Total Length Identification Flags Fragment Offset Time to live Protocol Header Checksum Source IP Address Destination IP Address IP Options (if any) Padding Data
IP Address & Forwarding n n n Based on destination address (32 bits!) n Not source (why is it there? ) n May change (or fail) somewhere along path Forwarding is “hop by hop” Address should be “where” something is n n Route is “how to get there” n n an interface of a host (can have lots) IP has several roles, conflated Routing Hint, Interface Id, Part of Flow State… … Computed seperately, continuously and asynchronously Names (see later) are “what” something is
Two components of routing n Control component n n Decides where the packets will go Use a set of routing protocols (e. g. OSPF, BGP) to collect information and produce a “forwarding table” “Control plane” routes Routing “daemon” collect routing info and maintain routing DB Forwarding table Forwarding component n n Moving packets from input to output ports according to forwarding table and packet header “Forwarding plane” packets Forwarding algorithm and mechanism kernel
Address Matching n Packet forwarding requires n Address matching Followed by table lookup of output port n Moving the packet through the router (from input port to output port) This involves scheduling, queueing, design of switch fabric etc, conventional aspects of switch design n Address matching n Exact matching e. g. bridge forwarding, DECnet, OSI/CLNP… n Longest prefix match – “best matching” IP networks
Exact match n n Easier Software approach: n n n Binary search Hash function Hardware: Content Addressable Memory (CAM)
Longest prefix match n n IP addresses are assigned in a manner that reflect network topology Address aggregation: group destinations with the same prefix together if they exit the same output port Therefore, longer prefixes tend to be announced by customers ISPs who are closer to the destination, whereas provider ISPs tend to announce aggregated addresses Hence a route to the longest prefix match is preferred
Example to show why “longest prefix match” is better Forwarding table ISP B (provider of ISP A) BGP route advertisement for 1. 2. 3/24 Peer relationship ISP C (provider of ISP A) Forwarding table 1. 2. 3/24 1. 2. 3. 123/26 BGP route advertisement for 1. 2. 3. 123/26 ISP A Subnet 1. 2. 3. 123/26 Longer prefix is a better route!
Example n Each entry in forwarding table has address + prefix e. g. Longest match address: 11001111 01011100 0000 10000111 mask: 11111111 address: 11001111 01011100 00000000 mask: 11111111 00000000 address: 11001111 01011100 00000000 mask: 11111111 11100000 11001111 01011100 0000 10000111 matches with all three entries
How to do Longest Prefix Match n n Not as easy as exact match Approaches: n n n Create a data structure for doing LPM Convert the problem into a form so that we can do binary search Reduce the problem to a sequence of exact match problems which we can apply hashing Optimization based on distribution of prefix lengths Combine software and hardware techniques
Algorithms There is an entire industry of algorithms: n Binary search among all prefixes in forwarding table Perlman’s book, 13. 4 Lampson et al “IP Lookups using Multiway and Multicolumn Search”, IEEE Infocom 1998 n Trie: bit-by-bit match Perlman’s book, 13. 3 n Binary search based on prefix length Perlman’s book, 13. 3. 3 Waldvogel et al “Scalable High Speed IP Routing Lookups”, Sigcomm 1997
But this is all going wrong! Why? n Not enough bits -> NATs… n n n Four M’s (historical order) n n n Multicast Mobility Multihoming Multipath Security and Social Scale n n n NAT Traversal, Stateful browser/server end is URL + Persistent HTTP state + cookie! Unsolicited traffic Byzantine (v. selfish or rational or altruistic) Despite original ARPANET packet radio n n And multicast since 1988, Hierarchy is wrong
So Ipng effort started in 1992 n See course web site for papers! n n n Specification of desiderata Led to a set of competing efforts Look at SIP & PIP n n n Represent extremes of CS (SIP) & Telco (PIP) SIP from PARC looks XNS n n Just ip with more address bits PIP looks VC/ATM ish… n Qo. S, fancy routing options
Eventually, converged on IPv 6 n n n Committee design (SIP/PIP/Novell) Overtaken by reality ? Four M’s (current order) n n n New requirements Receiver control of input n n n Multihoming - killing aggregation Mobility (smart phones roaming and receiving IP) Multipath (load balance, but how to id sub-flow) Multicast - sidelined? New kinds of bad guys Authentic addresses (HIP) New content type (video “interest”)
Next week… n n n Monday 15 th oct: Two of you will be speaking (Ben Confino & Ionel Gog on Plutarch and Haggle, respectively) Wednesday 17 th oct: I’ll set the first essay and discuss topic #2 Remember the assignments for speaking are online if you want to look ahead…
- Jon iversen
- Kalijum heksa cijano ferat
- Nomenklatura neorganskih jedinjenja
- Cviter jon
- Jon von neumann
- Jon peterson scholarship
- Jon hill project
- Jon christian thaulow
- Death of an innocent jon krakauer
- Jon's list of "five things successful people always do."
- Martin pratt camden
- Talus alge
- Jon courson theology
- Jon kjær nielsen
- Sulfatni jon
- Nitrovanje benzena
- John ramsey first wife
- Almond trees poem
- Jon petter arntzen
- Quality evangelist
- Jon malinowski photography