Download 15-441 Socket Programming

15-441: Computer Networking
Lecture 26: Where do we go from
here?
Overview
•
•
•
•
Content is king
Billions of devices
The next billion users
“Nothing is permanent but change”
2
Named Data Networking
• In the beginning...
– First applications strictly focused on host-to-host
interprocess communication:
• Remote login, file transfer, ...
– Internet was built around this host-to-host model.
– Architecture is well-suited for communication between pairs
of stationary hosts.
• ... while today
– Vast majority of Internet usage is data retrieval and service
access.
– Users care about the content and are oblivious to location.
They are often oblivious as to delivery time:
• Fetching headlines from CNN, videos from YouTube, TV from Tivo
• Accessing a bank account at www.bank.com.
3
To the beginning...
• What if you could re-architect the way
“bulk” data transfer applications worked
•
•
•
•
HTTP
FTP
Email
etc.
• ... knowing what we know now?
4
Google…
Biggest content source
Third largest ISP
Level(3)
Global
Crossing
Google
5
source: ‘ATLAS’ Internet Observatory 2009 Annual Report’, C. Labovitz et.al.
1995 - 2007:
Textbook Internet
2009:
Rise of the
Hyper Giants
6
source: ‘ATLAS’ Internet Observatory 2009 Annual Report’, C. Labovitz et.al.
What does the network look like…
ISP
ISP
7
What should the network look like…
ISP
ISP
8
Communication vs. Distribution
9
Overview
•
•
•
•
Content is king
Billions of devices
The next billion users
“Nothing is permanent but change”
10
Sensor Networks – Smart Devices
• First introduced in late 90’s by groups at
UCB/UCLA/USC
• Small, resource limited devices
• CPU, disk, power, bandwidth, etc.
• Simple scalar sensors – temperature, motion
• Single domain of deployment
• farm, battlefield, bridge, rain forest
• for a targeted task
• find the tanks, count the birds, monitor the bridge
• Ad-hoc wireless network
11
Sensor Example – Smart-Dust
• Hardware
•
•
•
•
•
•
UCB motes
4 MHz CPU
4 kB data RAM
128 kB code
50 kb/sec 917 Mhz radio
Sensors: light, temp.,
• Sound, etc.,
• And a battery.
12
Sensors, Power and Radios
• Limited battery life drives most goals
• Radio is most energy-expensive part.
• 800 instructions per bit. 200,000
instructions per packet. (!)
• That’s about one message per second for
~2 months if no CPU.
• Listening is expensive too. :(
13
Sensor Nets Goals
• Replace communication with computation
• Turn off radio receiver as often as possible
• Keep little state (limited memory).
14
Power
• Which uses less power?
• Direct sensor  base station Tx
• Total Tx power: distance^2
• Sensor  sensor  sensor  base station?
• Total Tx power: n * (distance/n) ^2 =~ d^2 / n
• Why? Radios are omnidirectional, but only one direction matters.
Multi-hop approximates directionality.
• Power savings often makes up for multi-hop capacity
• These devices are *very* power constrained!
15
Example: Aggregation
• Find average temperature in GHC 8th floor.
• Naïve: Flood query, let a collection point compute avg.
• Huge overload near the CP. Lots of loss, and local nodes use
lots of energy!
• Better:
• Take local avg. first, & forward that.
• Send average temp + # of samples
• Aggregation is the key to scaling these nets.
• The challenge: How to aggregate.
• How long to wait?
• How to aggregate complex queries?
• How to program?
16
Overview
•
•
•
•
Content is king
Billions of devices
The next billion users
“Nothing is permanent but change”
17
Example Routing Problem
2
Internet
City
bike
3
1
Village
18
Unstated Internet Assumptions
• Some path exists between endpoints
• Routing finds (single) “best” existing route
• E2E RTT is not very large
• Max of few seconds
• Window-based flow/cong ctl. work well
• E2E reliability works well
• Requires low loss rates
• Packets are the right abstraction
• Routers don’t modify packets much
• Basic IP processing
19
New Challenges
• Very large E2E delay
• Propagation delay = seconds to minutes
• Disconnected situations can make delay worse
• Intermittent and scheduled links
• Disconnection may not be due to failure (e.g.
LEO satellite)
• Retransmission may be expensive
• Many specialized networks won’t/can’t run
IP
20
What about TCP?
• Reliable in-order delivery streams
• Delay sensitive [6 timers]:
• connection establishment, retransmit, persist,
delayed-ACK, FIN-WAIT, (keep-alive)
• Three control loops:
• Flow and congestion control, loss recovery
• Requires duplex-capable environment
• Connection establishment and tear-down
21
Disruption Tolerant Networks
22
Disruption Tolerant Networks
23
Routing?
Village 2
City
bike (data mule)
intermittent high capacity
Geo satellite
medium/low capacity
dial-up link
low capacity
bandwidth
Village 1
time (days)
bike
satellite
phone
Connectivity: Village 1 – City
24
Overview
•
•
•
•
Content is king
Billions of devices
The next billion users
“Nothing is permanent but change”
25
Other Issues
•
•
•
•
Security
Mobility as the common case
Clouds and replicated services
Evolution support…
26
Now for a message
from the sponsors…
• Interested in this type of stuff?
• Networking group often takes students during
the semester or summer
• Stop by office hours or email to chat
27