1 Preliminaries, Protocols, Preview Priorities, and Principles EE122 Fall 2011 Scott Shenker Materials with thanks to Jennifer Rexford, Ion Stoica, Vern Paxson and other colleagues at Princeton and UC Berkeley

2 Announcements TAs gave me brutal feedback –Need to cover more of the basics –Goal today: fluency in packets, probability, terminology Homework #1 Out –This is about learning, not evaluation –But please try to do it on your own first –Will entertain questions in office hours and in class Today’s lecture will spill over into next week There is a special place in hell reserved for the designer of Powerpoint’s animation package…..

Outline Preliminaries –Life of a packet, types of delay, Little’s Law, etc. –Protocols –Client-server architectures Preview of homework assignment –Clear up any misunderstanding Design priorities (probably today) –Clark’s 1988 paper Design Principles (I hope not today) –The good stuff….. 3

Preliminaries

Nodes and Links Link: transmission technology –Twisted pair, optical, radio, whatever Node: computational devices on end of links –Host: general-purpose computer –Network node: switch or router 5 Node Link

Properties of Links Latency (delay) –Propagation time for data sent along the link –Corresponds to the “length” of the link Bandwidth (capacity) –Amount of data sent (or received) per unit time –Corresponds to the “width” of the link Bandwidth-delay product: (BDP) –Amount of data that can be “in flight” at any time –Propagation delay × bits/time = total bits in link 6 bandwidth latency delay x bandwidth

Examples of Bandwidth-Delay Same city over slow link: –B~100mbps –L~.1msec –BDP ~ 10000bits ~ 1.25MBytes Cross-country over fast link: –B~10Gbps –L~10msec –BDP ~ 10 8 bits ~ 12.5GBytes Another example where the Internet has to be prepared for a wide range of conditions! 7

Examples of Transmission Times 1500 byte packet over 14.4k modem: ~1 sec 1500 byte packet over 10Gbps link: ~10 -6 sec 8

Utilization Fraction of time link is busy transmitting –Often denoted by ρ Ratio of arrival rate to bandwidth –Arrival: A bits/sec on average –Utilization = A/B 9

Packets Payload (Body) –Data being transferred Header –Instructions to the network for how to handle packet –Think of the header as an interface! 10

The Lifecycle of Packets 11 Excess Packets Stored in Buffer Packet Arriving at Switch Packet Being Transmitted Packet Buffer Link Packet Currently Being Transmitted

The Delays of Their Lives 12 Queueing Delay Transmission Delay Propagation Delay is how long it takes to reach the next switch after transmission Round-Trip Time (RTT) is the time it takes a packet to reach the destination and the response to return to the sender

Review of Networking Delays Propagation delay: latency –Time spent in traversing the link –“speed of propagation” delay Transmission delay: –Time spent being transmitted –Ratio of packet size to bandwidth Queueing delay: –Time spent waiting in queue –Ratio of total packet bits ahead in queue to bandwidth Roundtrip delay (RTT) –Total time for a packet to reach destination and a response to return to the sender 13

Trends Propagation delay? –No change Transmission delay? –Getting smaller! Queueing delay? –Usually smaller How does this affect applications? –CDNs work very hard to move data near clients –Reduces backbone bandwidth requirements –But also decreases latency –Google: time is money! 14

Queueing Delay Does not happen if packets are evenly spaced –And arrival rate is less than service rate 15

Smooth Arrivals = No Queueing Delays 16

Queueing Delay Does not happen if packets are evenly spaced –And arrival rate is less than service rate Queueing delay caused by “packet interference” –Burstiness of arrival schedule –Variations in packet lengths 17

Bursty Arrivals = Queueing Delays 18 There is substantial queueing delay even though link is underutilized

Queueing Delay Does not happen if packets are evenly spaced –And arrival rate is less than service rate Queueing delay caused by “packet interference” –Burstiness of arrival schedule –Variations in packet lengths Made worse at high load –Less “idle time” to absorb bursts –Think about traffic jams in rush hour…. 19

Jitter Difference between minimum and maximal delay Latency plays no role in jitter –Nor does transmission delay for same sized packets Jitter typically just differences in queueing delay Why might an application care about jitter? 20

Packet Losses: Buffers Full 21

Packet Losses: Corruption 22

Reviewing Queueing: java applet 23

Basic Queueing Theory Terminology Arrival process: how packets arrive –Average rate A –Peak rate P Service process: transmission times –Average transmission time –For networks, function of packet size W: average time packets wait in the queue –W for “waiting time” L: average number of packets waiting in the queue –L for “length of queue” Two different quantities 24

Little’s Law (1961) L = A x W Compute L: count packets in queue every second –How often does a single packet get counted? W times Could compute L differently –On average, every packet will be counted W times –The average arrival rate determines how frequently this total queue occupancy should be added to the total Why do you care? –Easy to compute L, harder to compute W 25

Statistical Multiplexing

Data Rate 1 Data Rate 2 Data Rate 3 Three Flows with Bursty Arrivals Time Capacity

Data Rate 1 Data Rate 2 Data Rate 3 When Each Flow Gets 1/3 rd of Capacity Time Frequent Overloading

When Flows Share Total Capacity Time No Overloading Statistical multiplexing relies on the assumption that not all flows burst at the same time. Very similar to insurance, and has same failure case

Assume time divided into frames –Frames divided into slots Flows generate packets during each frame –Peak number of packets/frame P –Average number of packets/frame A Single flow: must allocate P slots to avoid drops –But P might be much bigger than A –Very wasteful! Use the “Law of Large Numbers”…. Another Take on “Stat Mux” 30 Frame Slots

Law of Large Numbers (~1713) Consider any probability distribution –Can be highly variable, such as varying from 0 to P Take N samples from probability distribution –In this case, one set of packets from each flow Thm: the sum of the samples is very close to N×A –And gets percentage-wise closer as N increases Sharing between many flows (high aggregation), means that you only need to allocate slightly more than average A slots per frame. –Sharing smooths out variations 31

If you still don’t understand stat mux Come to office hours and I can try another explanation… 32

Protocols

34 What Is A Protocol? Protocol: an agreement on how to communicate –To exchange data –To coordinate sharing of resources Protocols specifies syntax and semantics Syntax: how protocol is structured –Format, order messages are sent and received Semantics: what these bits mean –How to respond to various messages, events, etc.

Examples of Protocols in Life Asking a question in class Turn-taking in conversations –Pause is a signal for the next person’s response –When do these rules break? Boarding and exiting an airplane –Not all countries have the same protocol! Answering the phone –Starting with hello as signal for other party to talk –Other party identifies themselves, then conversation Teenagers (an example of unreliable transport!) –Information and money only flow in one direction 35

36 Example: HyperText Transfer Protocol GET /courses/archive/spring06/cos461/ HTTP/1.1 Host: User-Agent: Mozilla/4.03 CRLF HTTP/ OK Date: Mon, 6 Feb :09:03 GMT Server: Netscape-Enterprise/3.5.1 Last-Modified: Mon, 6 Feb :12:23 GMT Content-Length: 21 CRLF Site under construction Request Response

37 Example: IP Packet 4-bit Version 4-bit Header Length 8-bit Type of Service (TOS) 16-bit Total Length (Bytes) 16-bit Identification 3-bit Flags 13-bit Fragment Offset 8-bit Time to Live (TTL) 8-bit Protocol 16-bit Header Checksum 32-bit Source IP Address 32-bit Destination IP Address Options (if any) Payload 20-byteheader

Communication service –Ordered, reliable byte stream Both data transfer and resource sharing –Packet exchanges to make sure data arrives –Congestion control to share bandwidth fairly with others 38 Example: Transmission Control Protocol

39 Protocol Standardization All hosts must follow same protocol –Very small modifications can make a big difference –Or prevent it from working altogether –Cisco bug compatible! This is why we have standards –Can have multiple implementations of protocol Internet Engineering Task Force –Based on working groups that focus on specific issues –Produces “Request For Comments” (RFCs) –IETF Web site is –RFCs archived at

40 Internet Motto We reject kings, presidents, and voting. We believe in rough consensus and running code.” David Clark

Alternative to Standardization? Have one implementation used by everyone Open-source projects –Which has had more impact, Linux or POSIX? Or just sole-sourced implementation –Skype, many P2P implementations, etc. 41

Clients and Servers

43 Many Different Hosts on Internet Internet Also known as a “host”…

44 Clients and Servers Client program –Running on end host –Requests service –E.g., Web browser GET /index.html

45 Clients and Servers Client program –Running on end host –Requests service –E.g., Web browser Server program –Running on end host –Provides service –E.g., Web server GET /index.html “Site under construction”

46 Client-Server Communication Client “sometimes on” –Initiates a request to the server when interested –E.g., Web browser on your laptop or cell phone –Doesn’t communicate directly with other clients –Needs to know the server’s address Server is “always on” –Services requests from many client hosts –E.g., Web server for the Web site –Doesn’t initiate contact with the clients –Needs a fixed, well-known address

47 Peer-to-Peer Designs No always-on server at the center of it all –Hosts can come and go, and change addresses –Hosts may have a different address each time Example: peer-to-peer file sharing –All hosts are both servers and clients! –Scalability by harnessing millions of peers –“self-scaling” Not just for file sharing! –This is how many datacenter applications are built –Better reliability, scalability, less management… oSound familiar?

48 5 Minute Break Questions Before We Proceed?

Internet Design Goals

Internet Design Goals (Clark ‘88) Connect existing networks Robust in face of failures Support multiple types of delivery services Accommodate a variety of networks Allow distributed management Easy host attachment Cost effective Allow resource accountability 50

Connect Existing Networks Internet (e.g., IP) should be designed such that all current networks could support IP. 51

Robust As long as the network is not partitioned, two endpoints should be able to communicate Failures (excepting network partition) should not interfere with endpoint semantics Very successful, not clear how relevant now Second notion of robustness is underappreciated 52

53 Types of Delivery Services Use of the term “communication services” already implied an application-neutral network Built lowest common denominator service –Allow end-based protocols to provide better service Example: recognition that TCP wasn’t needed (or wanted) by some applications –Separated TCP from IP, and introduced UDP

54 Variety of Networks Incredibly successful! –Minimal requirements on networks –No need for reliability, in-order, fixed size packets, etc. –A result of aiming for lowest common denominator IP over everything –Then: ARPANET, X.25, DARPA satellite network.. –Now: ATM, SONET, WDM…

Decentralized Management Both a curse and a blessing –Important for easy deployment –Makes management hard today 55

56 Host Attachment Clark observes that cost of host attachment may be higher because hosts have to be smart But the administrative cost of adding hosts is very low, which is probably more important

Cost Effective Cheaper than telephone network But much more expensive than circuit switching Perhaps it is cheap where it counts (low-end) and more expensive for those who can pay…. 57

Resource Accountability Failure! 58

59 Internet Motto We reject kings, presidents, and voting. We believe in rough consensus and running code.” David Clark

Real Goals Build something that works! Connect existing networks Robust in face of failures Support multiple types of delivery services Accommodate a variety of networks Allow distributed management Easy host attachment Cost effective Allow resource accountability 60

61 Questions to think about…. What priorities would a commercial design have? What would the resulting design look like? What goals are missing from this list? Which goals led to the success of the Internet?

Internet Design Principles

The Networking Dilemma Many different networking technologies Many different network applications How do you prevent incompatibilities? 63

64 The Problem Re-implement every application for every technology? No! But how does the Internet design avoid this? Skype SSHNFS Radio Coaxial cable Fiber optic Application Transmission Media HTTP

65 Solution: Intermediate Layers Introduce intermediate layers that provide set of abstractions for various network functionality and technologies –A new app/media implemented only once –Variation on “add another level of indirection” Skype SSHNFS Packet radio Coaxial cable Fiber optic Application Transmission Media HTTP Intermediate layers

66 Network Architecture Architecture is not the implementation itself Architecture is how to organize/structure the elements of the system and their implementation What interfaces are supported? –Using what sort of abstractions Where functionality is implemented? –The modular design of the network

67 Computer System Modularity Partition system into modules & abstractions: Well-defined interfaces give flexibility –Hides implementation - can be freely changed –Extend functionality of system by adding new modules E.g., libraries encapsulating set of functionality E.g., programming language + compiler abstracts away how the particular CPU works …

Computer System Modularity (cnt’d) Well-defined interfaces hide information –Isolate assumptions –Present high-level abstractions But can impair performance! Ease of implementation vs worse performance 68

69 Network System Modularity Like software modularity, but: Implementation distributed across many machines (routers and hosts) Must decide: –How to break system into modules oLayering –Where modules are implemented oEnd-to-End Principle –Where state is stored oFate-sharing We will address these choices in turn

Remember that slide! The relationship between architectural principles and architectural decisions is crucial to understand 70