CS 2200 Lecture 25 TCPIP

1
CS 2200 Lecture 25TCP/IP

• (Lectures based on the work of Jay Brockman,
  Sharon Hu, Randy Katz, Peter Kogge, Bill Leahy,
  Ken MacKenzie, Richard Murphy, and Michael
  Niemier)

2
A disk, pictorially

• When accessing data we read or write to a sector
• All sectors the same size, outer tracks just less
  dense
• To read or write, moveable arm with read/write
  head moves over each surface
• Cylinder all tracks under the arms at a given
  point on all surfaces
• To read or write
• Disk controller moves arm over proper track a
  seek
• The time to move is called the seek time
• When sector found, data is transferred

3
The speed of light? No.

• Time required for a requested track sector to
  rotate under the read/write head is called the
  rotation latency or rotational delay
• Involves mechanical components on the order of
  milliseconds
• i.e. were no longer moving at the speed of light
  like in our CPU!
• Time required to actually write or read data is
  called the transfer time
• (a function of block size, rotation speed,
  recording density on a track, and speed of the
  electronics connecting the disk to the computer)

4
Disk odds n ends

• Often transfer time is a very small portion of a
  full access
• Its possible to use techniques (discussed in
  caches) to help reduce disk overhead. Any
  thoughts?
• To help reduce complexity theres usually
  additional HW called a disk controller
• Disk controller helps manage disk accesses
• but also adds more overhead controller time
• (Can also have a queuing delay)
• (Time spent waiting for a disk to become free if
  its already in use for another access)

5
Example average disk access time

• What is the average time to read or write a
  512-byte sector for a typical disk?
• The average seek time is given to be 9 ms
• The transfer rate is 4 MB per second
• The disk rotates at 7200 RPM
• The controller overhead is 1 ms
• The disk is currently idle before any requests
  are made (so there is no queuing delay)
• Average disk access time average seek time
  average rotational delay transfer time
  controller overhead

6
(Recall) Interrupts/Exceptions/Traps protection
I/O (kernel) space
a loop
user space
PC (mem. addr.)
a system call
kernel space
an interrupt
time
7
(Recall) Process States
New
Terminated
Ready
Running
A longer example using these states later on in
lecture
Waiting
8
DMA
N
Processor tells controller to make DMA
transfer. Assume disk to memory. (Includes N
number of bytes)
9
Arbitration

• DMA implies multiple owners of the bus
• must decide who owns the bus from cycle to cycle
• Arbitration
• Daisy chain
• Centralized parallel arbitration
• Distributed arbitration by self selection
• Distributed arbitration by collision detection
• (see board for detailed examples and pictures)

10
Physical Intuition Check

• If we read surface X, track Y, sector Z, whats
  the easiest next thing to read?
• Remember, the disk is actually rotating
• If we want to read all the sectors on the whole
  disk from 0,0,0 to the end, what order is fastest
  to read them?

11
Disk as one big array

• Interleave sectors, then surfaces, then tracks
• To store files, we need to allocate space out of
  this big array.
• Would like to have fast
• sequential access
• random access
• allocation/deallocation time
• Also
• nice to be able to grow files
• nice not to waste space
• Data structures?

181.6GB
0
12
Allocation Strategies

• 1. Fixed contiguous regions
• Regions (a.k.a. possible choices)
• One track
• One cylinder
• Contiguous range of cylinders
• Characteristics
• Data structures (on the disk at a well-known
  place)
• Free list bit map of free cylinders
• Directory mapping table (file name to cylinder
  address)
• SA and RA quick allocation is quick as well
• Cannot grow file size above allocation

-(
13
Allocation Strategies

• 2. Contiguous regions with overflow areas
• Secondary area for growth (also contiguous)
• Characteristics
• RA requires some computation
• SA as fast as 1.

14
Allocation Strategies

• 3. Linked allocation
• File divided into (sector sized) blocks
• Each block points to next one disk becomes a
  giant linked list!
• Characteristics
• Data structures (on the disk at a well-known
  place)
• free-list (a linked list) mapping table (file
  name to starting disk block)
• SA slow due to seek time RA pointer chasing
• Growth easy allocation is expensive
• (We have to find free boxes)

15
Allocation Strategies

• 4. File Allocation Table (FAT) MS-DOS
• At the beginning of each partition, a table that
  contains one entry for each disk block in that
  partition (0 free)
• A file occupies a number of entries in the FAT
• Each entry points to the next entry in the FAT
  for SA (-1 the last block)
• Characteristics
• FAT is the data structure efficient for
  allocation less chance of screwing up as in (3)
• SA requires FAT lookup (can be alleviated by
  caching the FAT) i.e. think multiple disk
  access
• RA requires some computation
• Limit on size of partition (this is BAD!) growth
  easy

16
Allocation Strategies

• 5. Indexed Allocation
• File represented by an index block (on the disk),
  a table of disk blocks for the file
• Characteristics
• Data structures
• Mapping table (file name to index block)
• Free list (can be a bit map of available disk
  blocks)
• Limit on the size of the file
• Wasted space
• See board

17
Allocation Strategies

• 6. Multilevel Indexed Allocation
• File represented by an index block (on the disk),
  an indirection table of disk blocks for that file
  
• One level indirection
• Two level indirection
• Triple indirection
• See next slide/board for picture

18
Multilevel Indexed Allocation
19
Allocation Strategies

• 7. Hybrid (BSD Unix)
• Combination of (5) and (6)
• Each file represented by an i-node (index node)
• Index to first n disk blocks, plus
• A single indirect index, a double indirect index,
  a triple indirect index
• Characteristics
• SA requires i-node reference
• overhead reduced by in-memory cache
• RA requires some computation but much quicker
  than (3)
• Growth is easy
• Allocation overhead same as (3)

20
Unix Inode
(see board talk) Q why is this a good strategy?
21
Disk Scheduling

• Algorithm Objective
• Minimize seek time
• Assumptions (for this example)
• Single disk module
• Requests for equal sized blocks
• Random distribution of locations on disk
• One movable arm
• Seek time proportional to tracks crossed
• No delay due to controller
• Read/write times are equal

22
FCFS
23
SSTF
24
SCAN
25
C-Scan
26
C-Look
27
Policies

• N-Step SCAN
• Two queues
• active (N requests or less)
• latent
• Service active queue
• When no more in active, transfer N requests from
  latent to active
• Leads to lower variance compared to SCAN
• Worse than SCAN for mean waiting time

28
Algorithm Selection

• SSTF (Shortest Seek Time First) is common. Better
  than FCFS.
• If load is heavy SCAN and C-SCAN best because
  less likely to have starvation problems
• We could calculate an optimum for any series of
  requests but costly
• Depends on number type of requests
• e.g. Imagine we only have one request pending
• Also depends on file layout
• Recommend a modular scheduling algorithm that can
  be changed.

29
Typical Question

• Suppose a disk drive has 5000 cylinders, numbered
  from 0 to 4999. Currently at cylinder 143 and
  previous request was at 125. Queue (in FIFO
  order) is
• 86, 1470, 913, 1774, 948, 1509, 1022, 1750, 130
• Starting from current position what is total
  distance moved (in cyclinders) the disk head
  moves to satisfy all requests
• Using FCFS, SSTF, SCAN, LOOK, C-SCAN

30
FCFS

• 143 - 86 57
• 86 - 1470 1384
• 1470 - 913 557
• 913 - 1774 861
• 1774 - 948 826
• 948 - 1509 561
• 1509 - 1022 487
• 1022 - 1750 728
• 1750 - 130 1620
• 7081 Cylinderslt-- Answer

31
Flynns taxonomy

• Single instruction stream, single data stream
  (SISD)
• Essentially, this is a uniprocessor
• Single instruction stream, multiple data streams
  (SIMD)
• Same instruction executed by multiple processors
  with different data streams
• Each processor has own data memory, but 1
  instruction memory and control processor to
  fetch/dispatch instructions

32
Flynns Taxonomy

• Multiple instruction streams, single data streams
  (MISD)
• Can anyone think of a good application for this
  machine?
• Multiple instruction streams, multiple data
  streams (MIMD)
• Each processor fetches its own instructions and
  operates on its own data

33
Sharing Data (another view)
Uniform Memory Access - UMA
Memory
Symmetric Multiprocessor SMP
34
Sharing Data (another view)
Non-Uniform Memory Access - NUMA
35
Communicating between nodes

• One way to communicate b/t processors treats
  physically separate memories as 1 big memory
• (i.e. 1 big logically shared address space)
• Any processor can make a memory reference to any
  memory location even if its at a different node
• Machines are called distributed shared
  memory(DSM)
• Same physical address on two processors refers to
  the same one location in memory
• Another method involves private address spaces
• Memories are logically disjoint cannot be
  addressed be a remote processor
• Same physical address on two processors refers to
  two different locations in memory
• These are multicomputers

36
Multicomputer
Proc Cache A
Proc Cache B
interconnect
memory
memory
37
But both can have a cache coherency problem
Cache A
Cache B
Read X Write X
Read X ...
Oops!
X 1
X 0
memory
X 0
38
Cache coherence protocols

• Directory Based
• Whether or not some physical memory location is
  shared or not is recorded in 1 central location
• Called the directory
• Snooping
• Every cache w/entries from centralized main
  memory also has a particular blocks sharing
  status
• No centralized state kept
• Caches connected to shared memory bus
• If there is bus traffic, caches check (or
  snoop) to see if they have the block being
  transferred on bus
• Main focus of upcoming discussion

39
Side note Snoopy Cache
CPU
CPU references check cache tags (as
usual) Cache misses filled from memory (as
usual) Other read/write on bus must check tags,
too, and possibly invalidate
State Tag Data
Bus
40
Write invalidate example

• Assumes neither cache had value/location X in it
  1st
• When 2nd miss by B occurs, CPU A responds with
  value canceling response from memory.
• Update Bs cache memory contents of X updated
• Typical and simple

41
Maintaining the cache coherency requirement

• Alternative to write invalidate update all
  cached copies of a data item when the item is
  written
• Called a write update/broadcast protocol
• One problem bandwidth issues could quickly get
  out of hand
• Solution track whether or not a word in the
  cache is shared (i.e. contained in another cache)
• If the word is not shared, theres no need to
  broadcast on a write

42
Write update example
(Shaded parts are different than before)

• Assumes neither cache had value/location X in it
  1st
• CPU and memory contents show value after
  processor and bus activity both completed
• When CPU A broadcasts the write, cache in CPU B
  and memory location X are updated

43
Advantages and disadvantages

• Shared memory good
• Compatibility with well-understood mechanisms in
  use in centralized multiprocessors used shared
  memory
• Its easy to program
• Especially if communication patterns are complex
• Easier just to do a load/store operation and not
  worry about where the data might be (i.e. on
  another node with DSM)
• But, you also take a big time performance hit
• Smaller messages are more efficient w/shared
  memory
• Might communicate via memory mapping instead of
  going through OS
• (like wed have to do for a remote procedure call)

44
Advantages and disadvantages

• Shared memory good (continued)
• Caching can be controlled by the hardware
• Reduces the frequency of remote communication by
  supporting automatic caching of all data
• Message-passing good
• The HW is lots simpler
• Especially by comparison with a scalable
  shared-memory implementation that supports
  coherent caching of data
• Communication is explicit
• Forces programmers/compiler writers to think
  about it and make it efficient
• This could be a bad thing too FYI

45
Specifics of snooping

• Normal cache tags can be used
• Existing valid bit makes it easy to invalidate
• What about read misses?
• Easy to handle too rely on snooping capability
• What about writes?
• Wed like to know if any other copies of the
  block are cached
• If theyre NOT, we can save bus bandwidth
• Can add extra bit of state to solve this problem
  state bit
• Tells us if block is shared, if we must generate
  an invalidate
• When write to a block in shared state happens,
  cache generates invalidation and marks block as
  private
• No other invalidations sent by that processor for
  that block

46
Specifics of snooping

• When invalidation sent, state of owners
  (processor with sole copy of cache block) cache
  block is changed from shared to unshared (or
  exclusive)
• If another processor later requests cache block,
  state must be made shared again
• Snooping cache also sees any misses
• Knows when exclusive cache block has been
  requested by another processor and state should
  be made shared

47
Specifics of snooping

• More overhead
• Every bus transaction would have to check
  cache-addr. tags
• Could easily overwhelm normal CPU cache accesses
• Solutions
• Duplicate the tags snooping/CPU accesses can go
  on in parallel
• Employ a multi-level cache with inclusion
• Everything in the L1 cache also in L2 snooping
  checks L2, CPU L1

48
Threads
Recall from board code, data, files shared No
process context switching

• Can be context switched more easily
• Registers and PC
• Not memory management
• Can run on different processors concurrently in
  an SMP
• Share CPU in a uniprocessor
• May (Will) require concurrency control
  programming like mutex locks.

This is why we talked about critical sections,
etc. 1st
49
Process Vs. Thread
P1
P2
user
PCB
PCB
kernel
Kernel code and data

• Two single-threaded applications on one machine

50
Process Vs. Thread
P1
P2
user
PCB
PCB
kernel
Kernel code and data

• P1 is multithreaded P2 is single-threaded
• Computational state (PC, regs, ) for each thread
• How different from process state?

51
Things to know?

• The reason threads are around?
• 2. Benefits of increased concurrency?
• 3. Why do we need software controlled "locks"
  (mutexes) of shared data?
• 4. How can we avoid potential deadlocks/race
  conditions.
• 5. What is meant by producer/consumer thread
  synchronization/communication using pthreads?
• 6. Why use a "while" loop around a
  pthread_cond_wait() call?
• 7. Why should we minimize lock scope (minimize
  the extent of code within a lock/unlock block)?
• 8. Do you have any control over thread
  scheduling?

52
See handwritten notes for more on mutex,
condition variables, and threads
53
3 kinds of networks

• Massively Parallel Processor (MPP) network
• Typically connects 1000s of nodes over a short
  distance
• Often banks of computers
• Used for high performance/scientific computing
• Local Area Network (LAN)
• Connects 100s of computers usually over a few kms
• Most traffic is 1-to-1 (between client and
  server)
• While MPP is over all nodes
• Used to connect workstations together (like in
  Fitz)
• Wide Area Network (WAN)
• Connects computers distributed throughout the
  world
• Used by the telecommunications industry

54
Performance parameters (see board)

• Bandwidth
• Maximum rate at which interconnection network can
  propagate data once a message is in the network
• Usually headers, overhead bits included in
  calculation
• Units are usually in megabits/second, not
  megabytes
• Sometimes see throughput
• Network bandwidth delivered to an application
• Time of Flight
• Time for 1st bit of message to arrive at receiver
• Includes delays of repeaters/switches length /
  m (speed of light) (m determines property of
  transmission material)
• Transmission Time
• Time required for message to pass through the
  network
• size of message divided by the bandwidth

55
Performance parameters (see board)

• Transport latency
• Time of flight transmission time
• Time message spends in interconnection network
• But not overhead of pulling out or pushing into
  the network
• Sender overhead
• Time for mP to inject a message into the
  interconnection network including both HW and SW
  components
• Receiver overhead
• Time for mP to pull a message out of
  interconnection network, including both HW and SW
  components
• So, total latency of a message is

56
Some more odds and ends

• Note from the example (with regard to longer
  distance)
• Time of flight dominates the total latency
  component
• Repeater delays would factor significantly into
  the equation
• Message transmission failure rates rise
  significantly
• Its possible to send other messages with no
  responses from previous ones
• If you have control of the network
• Can help increase network use by overlapping
  overheads and transport latencies
• Can simplify the total latency equation to
• Total latency Overhead (Message
  size/bandwidth)
• Leads to
• Effective bandwidth Message size/Total latency

57
Switched vs. shared
Node
Node
Node
Shared Media (Ethernet)
Node
Node
Switched Media (ATM)
(A. K. A. data switching interchanges,
multistage interconnection networks, interface
message processors)
Switch
Node
Node
58
Switch topology

• Switch topologyreally just a fancy term for
  describing
• How different nodes of a network can be connected
  together
• Many topologies have been proposed, researched,
  etc. only a few are actually used
• MPP designers usually the most creative
• Have used regular topologies to simplify
  packaging, scalability
• LANs and WANs more random
• Often a function of what equipment is around,
  distances, etc.
• Two common switching organizations
• Crossbar
• Allows any node to communicate with any other
  node with 1 pass through an interconnection
• Omega
• Uses less HW (n/2 log2n vs. n2 switches) more
  contention

59
Connection-Based vs. Connectionless

• Telephone operator sets up connection between
  the caller and the receiver
• Once the connection is established, conversation
  can continue for hours
• Share transmission lines over long distances by
  using switches to multiplex several conversations
  on the same lines
• Time division multiplexing divide B/W
  transmission line into a fixed number of slots,
  with each slot assigned to a conversation
• Problem lines busy based on number of
  conversations, not amount of information sent
• Advantage reserved bandwidth

(see board for ex.)
60
Routing Messages

• Shared Media
• Broadcast to everyone!
• Switched Media needs real routing. Options
• Source-based routing message specifies path to
  the destination (changes of direction)
• Virtual Circuit circuit established from source
  to destination, message picks the circuit to
  follow
• Destination-based routing message specifies
  destination, switch must pick the path
• deterministic always follow same path
• adaptive pick different paths to avoid
  congestion, failures
• randomized routing pick between several good
  paths to balance network load

61
Store and Forward vs. Cut-Through

• Store-and-forward policy each switch waits for
  the full packet to arrive in switch before
  sending to the next switch (good for WAN)
• Cut-through routing or worm hole routing switch
  examines the header, decides where to send the
  message, and then starts forwarding it
  immediately
• In worm hole routing, when head of message is
  blocked, message stays strung out over the
  network, potentially blocking other messages
  (needs only buffer the piece of the packet that
  is sent between switches).
• Cut through routing lets the tail continue when
  head is blocked, accordioning the whole message
  into a single switch. (Requires a buffer large
  enough to hold the largest packet).
• See board

62
Ethernet Evolution

• X_Base_Y
• X stands for the available media bandwidth
• Base stands for base band signaling on the medium
• Y stands for the maximum distance a station can
  be from the vampire tap (i.e. Length of Attach
  Unit Interface)

63
More detail

• In 100BaseT
• In 100 Mbps Ethernet (known as Fast Ethernet),
  there are three types of physical wiring that can
  carry signals
• 100BASE-T4 (four pairs of telephone twisted pair
  wire
• 100BASE-TX (two pairs of data grade twisted-pair
  wire)
• 100BASE-FX (a two-strand optical fiber cable)  
• This designation is an Institute of Electrical
  and Electronics Engineers shorthand identifier.
• The "100" in the media type designation refers to
  the transmission speed of 100 Mbps.
• The "BASE" refers to base band signaling
• i.e. only Ethernet signals are carried on the
  medium
• The "T4," "TX," and "FX" refer to the physical
  medium that carries the signal.
• (Through repeaters, media segments of different
  physical types can be used in the same system.)
• Source http//www.ggrego.com/glossary/glossary_n
  um.htm

64
A short summary

• This designation is an Institute of Electrical
  and Electronics Engineers (IEEE) shorthand
  identifier (i.e. X BASE Y)
• The "10" in the media type designation refers to
  the transmission speed of 10 Mbps.
• The "BASE" refers to base band signaling, which
  means that only Ethernet signals are carried on
  the medium.
• The "T" represents twisted-pair the "F"
  represents fiber optic cable
• and the "2", "5", and "36" refer to the coaxial
  cable segment length
• (the 185 meter length has been rounded up to "2"
  for 200).

65
Broadband vs. Baseband

• A baseband network has a single channel that is
  used for communication between stations. Ethernet
  specifications which use BASE in the name refer
  to baseband networks.
• BASE refers to BASE BAND signaling. Only
  Ethernet signals are carried on the medium
• A broadband network is much like cable
  television, where different services communicate
  across different frequencies on the same cable.
• Broadband communications would allow a Ethernet
  network to share the same physical cable as voice
  or video services. 10BROAD36 is an example of
  broadband networking.

66
Ethernet

• The various Ethernet specifications include a
  maximum distance
• What do we do if we want to go further?
• Repeater
• Hardware device used to extend a LAN
• Amplifies all signals on one segment of a LAN and
  transmits them to another
• Passes on whatever it receives (GIGO)
• Knows nothing of packets, addresses
• Any limit?

67
Repeaters
R1
R2
R3
68
Bridges

• We want to improve performance over that provided
  by a simple repeater
• Add functionality (i.e. more hardware)
• Bridge can detect if a frame is valid and then
  (and only then) pass it to next segment
• Bridge does not forward interference or other
  problems
• Computers connected over a bridged LAN don't know
  that they are communicating over a bridge

69
Bridges

• Typical bridge consists of conventional CPU,
  memory and two NIC's.
• Does more than just pass information from one
  segment to another
• A bridge can be constructed to
• Only pass valid frame if necessary
• Learn what is connected to network "on the fly"

70
Ethernet vs. Ethernet w/bridges
Node
Node
Node
Node
Node
Node
Node
Node
Node
Node
Node
Single Ethernet 1 packet at a time
Node
Node
Node
Node
Node
Node
Bridge
Bridge
Node
Node
Node
Node
Node
Multiple Ethernets Multiple packets at a time
71
Network Interface Card

• NIC
• Sits on the host station
• Allows a host to connect to a hub or a bridge
• Hub merely extends multiple segments into
  single LAN do not help with performance only 1
  message can transmit at a time
• If connected to a hub, then NIC has to use
  half-duplex mode of communication (i.e. it can
  only send or receive at a time)
• If connected to a bridge, then NIC (if it is
  smart) can use either half/full duplex mode
• Bridges learn Media Access Control (MAC) address
  and the speed of the NIC it is talking to.

72
Routers

• Routers
• Devices that connect LANs to WANs or WANs to WANs
• Resolve incompatible addresses (generally slower
  than bridges)
• Divides interconnection networks into smaller
  subnets which simplifies manageability and
  security
• Work much like bridges
• Pay attention to the upper network layer
  protocols
• (OSI layer 3) rather than physical layer (OSI
  layer 1) protocols.
• (This will make sense later)
• Will decide whether to forward a packet by
  looking at the protocol level addresses (for
  instance, TCP/IP addresses) rather than the MAC
  address.
• (This will make sense later)

73
Why we need protocols/layers

• Enable sharing the hardware network links
• Overcome sources of unreliability in the network
• Lost packets
• Temporary failure of an intervening routing node
• Mangled packets
• Reflections on the media, soft errors in
  communication hardware buffers, etc.
• Out of order delivery
• Packets of the same message routed via different
  intervening nodes leading to different latencies

74
Recall

• A protocol is the set of rules used to describe
  all of the hardware and (mostly) software
  operations used to send messages from Processor A
  to Processor B
• Common practice is to attach headers/trailers to
  the actual payload forming a packet or frame.

75
Protocol Family Concept
Message
Message
Message
76
Layering Advantages

• Layering allows functionally partitioning the
  responsibilities (similar to having procedures
  for modularity in writing programs)
• Allows easily integrating (plug and play) new
  modules at a particular layer without any changes
  to the other layers
• See the board
• Rigidity is only at the level of the interfaces
  between the layers, not in the implementation of
  these interfaces
• By specifying the interfaces judiciously
  inefficiencies can be avoided

77
ISO Model
7

• Interact with user e.g. mail, telnet, ftp

Presentation

• Char conv., echoing, format diffs endian-ness

6
Session

• Process to process comm. e.g. Unix sockets

5
Transport

• Packetizing, seq-num, retrans. e.g. TCP, UDP

4
Network

• Routing, routing tables e.g. IP

3

• Interface to physical media, error recovery e.g.
  retransmit on collision in Ethernet

Data Link
2

• Electrical and mechanical characteristics of
  physical media e.g. Ethernet

Physical
1
78
ISO Model Examples
7
User program

• FTP

Presentation
6
Session

• Sockets open/close/read/write interface

5
Kernel Software
Transport

• TCP reliable infinite-length stream

4
Network

• IP unreliable datagrams anywhere in
  world

3

• Ethernet unreliable datagrams on local segment

Data Link
2
Hardware

• 10baseT ethernet spec twisted pair w/RJ45s

Physical
1
79
Layering Summary

• Key to protocol families is that communication
  occurs logically at the same level of the
  protocol, called peer-to-peer,
• but is implemented via services at the next lower
  level
• Encapsulation carry higher level information
  within lower level envelope
• Fragmentation break packet into multiple smaller
  packets and reassemble
• Danger is each level increases latency if
  implemented as hierarchy (e.g., multiple check
  sums)

80
Layering SummaryTCP atop IP atop Ethernet

• Application sends message

• TCP breaks into 64KB segments, adds 20B header

• IP adds 20B header, sends to network

• If Ethernet, broken into 1500B packets with
  headers, trailers (24B)

• All Headers, trailers have length field,
  destination, ...

81
Techniques Protocols Use

• Sequencing for Out-of-Order Delivery
• Attach sequence number to packet, Maintain
  counter, If arriving packet is next sequence pass
  it on, If not save until correct point
• Sequencing to Eliminate Duplicate Packets
• Use sequence numbers to detect and discard
  duplicates
• Retransmitting Lost Packets
• Avoiding Replay Caused by Excessive Delay
• Protocols mark each session with unique Session
  ID
• Incorrect session ID causes packet to be
  discarded
• Flow Control to Prevent Data Overrun
• Use sliding window
• Mechanism to Avoid Network Congestion
• Switches detect congestion use packet loss to
  adjust packet rate

82
Naming and Name Resolution

• Within a system each process has an ID
• Across a network process have no knowledge of one
  another
• Generally processes are identified by
• lthost name, identifiergt
• Must have a system to resolve names
• 1. Every system can have file with complete
  listing of all other hosts Internet originally
  used this!
• 2. Distribute name information across network and
  have appropriate distribution and retrieval
  protocol
• This is Domain Name Server in
  use today

83
Domain Name Servers

• Imagine that a system wants to locate
• gaia.cc.gatech.edu
• The kernel will issue a request to a name
  server for the edu domain. This name server will
  be at a known address.
• The edu name server will issue the address where
  the gatech.edu name server is located.
• This name server is queried and it returns the
  address of cc.gatech.edu which when queried will
  return the Internet Address of gaia.cc.gatech.edu
  (130.207.9.18)

84
Next Hop Forwarding
This table is for switch 2.
85
Step oneDefine universal packet format
86
Step twoEncapsulate the universal packetsin
(any) local network frame format
Used to send msg. from 1 network to another (or
wi/the same)but we want a uniform standard.
Frame Header
Frame Data
Used to communicate within 1 network
87
Physical Network Connection
Router
Router facilitates communication
between networks
Individual Networks
Each cloud represents arbitrary network
technology LAN, WAN, ethernet, token ring, ATM,
etc.
88
Routers

• A router is
• a special-purpose computer dedicated to the task
  of interconnecting networks.
• A router can interconnect networks that use
  different technologies
• (including different media, physical addressing
  schemes or frame formats)

Router
89
Router operation

• Unpack IP packet from frame format of source
  network
• Perform routing decision
• Re-pack IP packet in frame format of the
  destination network
• (see board for demo packing, unpacking,
  repacking)

90
Reassembly

• Fragments are never reassembled until the final
  destination
• Why?
• Reduce amount of state information in routers.
  When packets arrive at a router they can simply
  be forwarded
• Allows routes to change dynamically. Intermediate
  reassembly would be problematic if all fragments
  didn't arrive.

91
Example
Source Host
Net 1
header 1
Router 1
Net 2
header 2
Router 2
Net 3
header 3
Destination Host
92
TCP/IP

• A number of different protocols have been
  developed to permit internetworking
• TCP/IP (actually a suite of protocols) was the
  first developed.
• Work began in 1970 (same time as LAN's were
  developed)
• Most of the development of TCP/IP was funded by
  the US Government (ARPA)

93
Layer upon layer upon layer...

• Layer 1 Physical
• Basic network hardware (same as ISO model Layer
  1)
• Layer 2 Network Interface
• How to organize data into frames and how to
  transmit over network (similar to ISO model Layer
  2)
• Layer 3 Internet
• Specify format of packets sent across the
  internet as well as forwarding mechanisms used by
  routers
• Layer 4 Transport
• Like ISO Layer 4 specifies how to ensure reliable
  transfer
• Layer 5 Application
• Corresponds to ISO Layers 6 and 7. Each Layer 5
  protocol specifies how one application uses an
  internet

94
IP Address Hierarchy

• Addresses are broken into a prefix and a suffix
  for routing efficiency
• The Prefix is uniquely assigned to an individual
  network.
• The Suffix is uniquely assigned to a host within
  a given network

1
1
2
Network 1
Network 2
3
3
5
95
Five Classes of IP Address
Primary Classes
96
Computingthe Class
(take a quiz) (then see the board)
97
Classes and Dotted Decimal

• Class
• A
• B
• C
• D
• E

• Range of Values
• 0 through 127
• 128 through 191
• 192 through 223
• 224 through 239
• 240 through 255

Does this mean there are 64 Class B networks?
Does this mean there are 32 Class C networks?
(on the board)
98
Division of the Address Space
Address Class
Bits in Prefix
Maximum Number of Networks
Bits in Suffix
Maximum Number of Hosts per Network
A B C
7 14 21
128 16384 2097152
24 16 8
16777216 65536 256
(on the board)
99
Special IP Address Summary
Prefix
Suffix
Type of Address
Purpose
All-0's
All-0's
This computer
Used during bootstarp
Network
All-0's
Network
Identifies a network
Network
all-1's
Directed broadcast
Broadcast on specified net
All-1's
All-1's
Limited broadcast
Broadcast on local net
127
Any
Loopback
Testing
Network
All-0's
Directed broadcast
Berkley broadcast
100
Routers and IP Addressing

• Each host has an address
• Each router has two (or more) addresses!
• Why?
• A router has connections to multiple physical
  networks
• Each IP address contains a prefix that specifies
  a physical network
• An IP address does not really identify a specific
  computer but rather a connection between a
  computer and a network.
• A computer with multiple network connections
  (e.g. a router) must be assigned an IP address
  for each connection

101
Example
Ethernet 131.108.0.0
Token Ring 223.240.129.0
131.108.99.5
223.240.129.2
223.240.129.17
78.0.0.17
WAN 78.0.0.0
Note!
(on the board)
102
Address Resolution Protocol

• IP addresses are virtual
• LAN/WAN hardware doesn't understand IP addresses
• Frame transmitted across a network must have
  hardware address of destination (in that network)
• Three basic mechanisms for resolving addresses

1. Address translation table Used primarily in
WAN's 2. Translation by mathematical function 3.
Distributed computation across network Protocol
addresses are abstractions Physical hardware does
not know how to locate a computer from its
protocol address Protocol address of next hop
must be must be translated to hardware address
103
Resolving Addresses

• 1. Address translation table
• Used primarily in WAN's
• 2. Translation by mathematical function
• 3. Distributed computation across network
• Protocol addresses are abstractions
• Physical hardware does not know how to locate a
  computer from its protocol addess
• Protocol address of next hop must be must be
  translated to hardware address

104
Address Resolution Protocol

• TCP/IP can use any of the three methods
• Table lookup usually used in a WAN
• Closed-form computation is used with configurable
  networks
• Message exchanged used in LAN's with static
  addressing
• To insure that all computers agree TCP/IP
  includes an Address Resolution Protocol
• Two types of messages are supported
• Request a hardware address given a protocol
  address
• Reply containing IP Address and hardware request

105
IP Addresses and Routing Table Entries
R1
R2
R3
Assume message with IP address
192.4.10.3 arrives at router R2
for each entry in table if(Mask Addr)
Dest forward to NextHop
(see board)
106
Routing Table Computation

• Routing tables are computed automatically
• Two basic approached are used
• Static routing
• Program runs when packet switch boots
• Advantages Simple with low network overhead
• Disadvantage Inflexible
• Dynamic routing
• Program builds routing table on boot and then as
  conditions change adjusts table
• Advantage Allows network to handle problems
  automatically