Transcript POTS and WANs - University of St. Thomas

CISC 370 - Class Today
•
•
•
•
•
The Lab
Project Schedule
Homework 7 and 8
Recap
POTS
4/2/2016
R. Smith - University of St Thomas - Minnesota
1
Upcoming Homework
• I’ll post some Chapter 12/13 homework soon
– Outline: April 23
• I’ll e-mail comments to your group
– Revised Outline (optional): Apr 30
– Papers: May 12
– Presentations: May 12 and 14
4/2/2016
R. Smith - University of St Thomas - Minnesota
2
A Lab Problem
• Someone plugged in the Firewall
• Caused ‘some’ machines to get Net 10
addresses
– The packets are discarded in most ‘real’ routers
– This causes alarm bells in some routers
– Fortunately we were not visited by angry IRT admins.
• I.E. Someone hooked up wires incorrectly
– The connection did NOT match the setup published in the Lab
• This was also NAUGHTY.
4/2/2016
R. Smith - University of St Thomas - Minnesota
3
Homework Review: Chapter 7
• 7.1: Mailbagging
– Good things: transmission efficiency, less intermediate storage
– Bad things: More work for endpoints (arguable)
• 7.3: One message is best
– SMTP sends one copy per server
– “Spam” vs “Velveeta”
• An old Internet discussion
• Spam = one message with many destinations
• Velveeta = many messages to separate destinations
4/2/2016
R. Smith - University of St Thomas - Minnesota
4
7.8: Tracing a Header
• http://courseweb.stthomas.edu/resmith/c/cisc370/c
9sp/headers.txt
• E-mail Forensics
– Every e-mail server prepends a ‘Received’ line
– Ideally the information is accurate
• In fact, it’s easy to forge
– Each server leaves earlier ‘Received’ lines undisturbed
– Typical forgery: modifies just the “From:” line.
– We can spot fakes by looking for inaccurate details
• Do “Received:” lines match the “From:” line?
• Do domain names match IP addresses?
• Do date/time stamps make sense?
4/2/2016
R. Smith - University of St Thomas - Minnesota
5
Homework: Chapter 8
• 8.1: a=net bits, b=host bits, c=#nets, d=#hosts, e=octet range
– Class A: (a) 8 bits, (b) 24 bits
• (c) first bit of the first octet in a class A address is 0 (leaving 7 bits), so 27 =
128 – 2 (0 and 127 are disallowed) = 126 networks,
• (d) 224 = 16,777,216 – 2 (host address cannot be all 0’s or all 1’s) =
16,777,214 hosts
• (e) range: 1 through 126
– Class B: (a) 16 bits, (b) 16 bits,
• (c) first two bits of the first octet in a class B address are 10 (leaving 14
bits), so 214 = 16,384 networks,
• (d) 224 = 65,536 – 2 (host address cannot be all 0’s or all 1’s) = 65,534 hosts,
(e) range: 128 through 191
– Class C: (a) 24 bits, (b) 8 bits,
• (c) first three bits in the first octet in a class C address are 110 (leaving 21
bits), so 221 = 2,097,152 networks,
• (d) 28 = 256 – 2 (host address cannot be all 0’s or all 1’s) = 254 hosts,
• (e) range: 192 through 223
4/2/2016
R. Smith - University of St Thomas - Minnesota
6
Chapter 8 homework, continued
• 8.5: Address 192.168.100; mask 255.255.192.0
– Class B-C “CIDR” address – no subnets, 214 hosts (16K)
– #0 bits = 14 – If taken as Class B:
• a) # Subnet bits = 16-15 = 2 bits => 4 subnets
• b) # Host bits = 14 => 16K hosts
• 8.8: Exhausting address spaces
–
–
–
–
–
31,557,600 seconds per year, per 10 seconds; per second
IPV4: easy answer = 232 => 4 billion or so => 1,360 years
IPV4: “book solution” = 221 “available” => 242 days
IPV6: based on book misprint = 2125 => 35 min
IPV6: for real = 2125 => 1030 years
• 8.9: Finding mask and gateway
– # of hosts on network
4/2/2016
R. Smith - University of St Thomas - Minnesota
7
The Plain Old Telephone System
• POTS
– Architecture (recap)
– SS-7
• WANs from the POTS folks
– X.25
– Frame Relay
– ATM
4/2/2016
R. Smith - University of St Thomas - Minnesota
8
Traditional POTS Architecture
• Provides analog connections to endpoints
– Digital features added atop analog voice-quality signals
– Digital emphasis has evolved in the past 10-15 years
• Grandly hierarchical
– “Routing” is based on the hierarchy
• Country code -> select the appropriate PTT
• 3 digit area code -> regional long distance exchange
• 3 digit exchange -> loop’s central office
• 4 digit subscriber number -> local loop wiring
– Routing within a region
• Ad-hoc but fixed
• Mostly relies on redundant connections to all destinations
4/2/2016
R. Smith - University of St Thomas - Minnesota
9
Telephone connection sequence
•
•
•
•
•
•
•
•
•
Both phones are on-hook, One goes off-hook
End office sends dial tone
Caller dials a number
Switch uses this as the ‘called address’
If called address is not busy, make it ring
Send ring tone to caller
If called phone goes off hook, connect the call
Turn off the ring signal
Continue the connection till a phone goes onhook
4/2/2016
R. Smith - University of St Thomas - Minnesota
10
Office-to-office connection
• Originating office finds a free connection on an
interoffice trunk
• Sends a request for a ‘digit register’ to receive
the called number
• Destination sends a ‘wink’ when it has a digit
register for originator to use
• Originator sends the number to the destination
office
• The destination connects to the end subscriber
loop, or continues through another office
4/2/2016
R. Smith - University of St Thomas - Minnesota
11
Signaling System 7 (SS7)
• Today, trunks use SS7 for control signaling
• Packet technology + POTS office architecture
– “Offices” are now called “switches”
– Highly redundant
• Supports modern capabilities
–
–
–
–
–
Phone numbers not tied to hardware (subscriber loop)
Phone numbers ‘roam’ geographically
Remote voice mail
Toll free numbers (800 etc)
Special charge numbers (900 etc)
4/2/2016
R. Smith - University of St Thomas - Minnesota
12
Elements of SS7
• These devices are deployed redundantly
• Service Switching Points (SSPs)
– Connect to subscribers – local loops
– Connect to STPs via SS7
– Sends queries to SCPs to find out how to route a call
• Service Transfer Points (STPs)
– A packet switch tailored to handle SS7
– Routes data based on phone numbers
– Firewalling traffic from ‘external’ networks
• Service Control Points (SCPs)
– Centralized databases
– Links particular phone numbers to particular subscribers
– Provides routing information for reaching subscribers
4/2/2016
R. Smith - University of St Thomas - Minnesota
13
WANs - The Telcos' parting attempts at
relevance
• They really are mired in an existing business
model and customer base
• Makes it hard for them to deal with the
changing data comm landscape
• You can almost see how modern services like
ATM reflect demands by particular (large)
customers with particular expectations
• Telcos still exist because they can meet these
demands and charge high tariffs for them.
4/2/2016
R. Smith - University of St Thomas - Minnesota
14
Classic WAN Lineup
• "Leased Lines" - dedicated point to point
connections (archaic!!)
• Most of these were a fixed (huge!) cost per
month
• Cost tied to distance of connection
– Analog - an ancient and relatively slow service (56K)
– Digital Data Service - a slightly less ancient and slow service
(56K)
– T-1 - the workhorse for early Internet sites 1.54M
– T-3 - something of an improvement: 44.7M
4/2/2016
R. Smith - University of St Thomas - Minnesota
15
Newer Services
• Frame Relay - more recent service
–
–
–
–
–
(talk more about it in a minute)
44.7M
Charge per month for the connecting port
Added charge per month for each virtual circuit's capacity
No extra charge for longer distances
• Synchro Optical Net (SONET) 51.4M to ...
– Standard designation for optical hardware connections
– “OC” numbers
• OC-1 (or STS-1) at 50Mb/sec
– thru
• OC-192 (STS-192) at 9.6Gb/sec
• STS-768 at 38 Gb/sec.. etc.
4/2/2016
R. Smith - University of St Thomas - Minnesota
16
"Switched Services”
•
•
•
•
gee, a choice of destinations!
Dial-up analog - the classic modem connection 56K
X.25 packet switching - now archaic 56K
ISDN – a first attempt at integrated ditigtal service:
– up to 1.54M
– cost per month plus connect time charge + long distance charges
• ADSL - something more contemporary, but aging:
– up to 9M
• Frame Relay - see, both switched and unswitched
• ATM - the Great White Hope of the telcos
– if this doesn't bring in business, they're history
– Pricing structure varies, but is not usually distance sensitive
4/2/2016
R. Smith - University of St Thomas - Minnesota
17
•Trade-offs between choices
• Cost structure: per link, per connection, per
packet, distance sensitive, etc.
• Switched vs unswitched
• Channels per physical link: all in one, or
multiplexed
• Reliability and flow control: network or
endpoint responsibility?
4/2/2016
R. Smith - University of St Thomas - Minnesota
18
X-25 Network Protocol
• Telco industry’s first - unsuccessful - attempt
to build a networking protocol
• Designed a "smart network“
• Misused the notion of a protocol stack
– used it to establish independence among protocol designers at
different levels – led to serious inefficiencies
– Flow control and error correction replicated at layers 2 and 3
4/2/2016
R. Smith - University of St Thomas - Minnesota
19
X.25 Architecture
• Telcos took as an article of faith that
connections are fundamental
– Embedded per-connection overhead in individual network
switches
– Personally, I implemented X.25 over the Arpanet backbone
without such foolishness and it worked fine.
– Flow control took some fine-tuning, but that worked, too.
• Services
– Cost per packet - I remember this; probably a link cost, too
– Multiple channels per link possible
– Switched and unswitched channels possible ('permanent'
virtual circuits)
4/2/2016
R. Smith - University of St Thomas - Minnesota
20
Frame Relay
• A "dumber network" than X.25
– closer to “end to end” Internet architecture concept
• WAN with unreliable datagrams and no flow
control
– Relies on end-to-end protocols like TCP to handle flow control
and error correction
– 'Smarter' than datagrams –
• retains order of transmission on a channel
– Stallings argues that this works because modern digital
transmission methods are more reliable than the analog
modem-based techniques
– Greatly increased network efficiency and reduced transmission
delays by eliminating "smart network" protocol overhead
4/2/2016
R. Smith - University of St Thomas - Minnesota
21
Protocol details
• Multiple channels –
– channel 0 for linking other channels to endpoints
• Each channel can have its own endpoint –
– either predefined or on a "per call" basis
– Like ‘virtual circuits’ on X.25
• Individual packets carry a channel number or
"Data Link Connection Identifier" (DLCI).
4/2/2016
R. Smith - University of St Thomas - Minnesota
22
Setting up a connection
• Initating host sends a SETUP packet - crosses the
network to the destination, delivered to destination
host.
• Destination host accepts by sending a CONNECT
packet - goes back to the initiating host.
• The SETUP/CONNECT protocol establishes a channel,
assigns a DLCI.
• When connection finished, send a RELEASE to other
end
• Other end responds with RELEASE COMPLETE
• No big deal - just different names for the same sort of
thing.
4/2/2016
R. Smith - University of St Thomas - Minnesota
23
Congestion control
• Not much.
• "Danger Will Robinson" bit –
– says that there's congestion in one direction or the other.
– "Forward/Backward Explicit Congestion Notification" FECN or BECN)
• "Sacrificial Lamb" bit –
– says this packet is a good one to discard if things are too congested.
– "Discard Eligibility" DE
• Implement multiple transmission rates, based on what
is paid for
– Committed Info Rate (CIR) - what's paid for
– Maximum Rate (MR) - what is accepted
– Access Rate –
• what the link accepts –
• excess past MR gets discarded
4/2/2016
R. Smith - University of St Thomas - Minnesota
24
•ATM or "Cell Relay“
• A "cell" is a "frame" only it's supposed to be
transmitted faster.
– Dumber and more efficient than X.25
– Cell sequence is preserved
• Basic Features
– Virtual channels
– Packet format/features
– Service categories
4/2/2016
R. Smith - University of St Thomas - Minnesota
25
Virtual paths and virtual channels
• Users see virtual channels as logical
connections
• Virtual paths are a network level property:
– represents a set of virutal channels with a common destination
–
– network handles them as an aggregated entity instead of
handling the channels individually
4/2/2016
R. Smith - University of St Thomas - Minnesota
26
Packet format
• Packet destination = virtual path + virtual
channel within path
• Payload type = user data vs system data,
– also includes info about congestion
– poor flow control again
• Sacrificial lamb bit - "Cell Loss Priority" (CLP)
• 8-bit checksum for the header
– since bit errors could cause pain to the network
4/2/2016
R. Smith - University of St Thomas - Minnesota
27
ATM Service categories
• or, "I'm a big customer and you'd better provide me the
category of service I want or I'm calling in the
competition."
– + Constant bit rate (CBR) - traditional connection service
– + Variable Bit Rat (VBR) - gives network more flexibility and lower cost
to the customer
– + Unspecified Bit Rate (UBR) - 'best effort' service - give it whatever
bandwidth is left over
– + Avaliable bit rate (ABR) - specifies a minimum cell rate required
(MCR) and a peak rate (PCR). Connects LANs across ATM
– + Guaranteed Frame Rate (GFR)
• - for connecting to Internet backbone. Has the ATM net understand
frame boundaries, so packets are discareded in "frame" sets
instead of individually, possibly from separate frames.
4/2/2016
R. Smith - University of St Thomas - Minnesota
28
Creative Commons License
This work is licensed under the Creative
Commons Attribution-Share Alike 3.0 United
States License. To view a copy of this license,
visit http://creativecommons.org/licenses/bysa/3.0/us/ or send a letter to Creative
Commons, 171 Second Street, Suite 300, San
Francisco, California, 94105, USA.
4/2/2016
R. Smith - University of St Thomas - Minnesota
29