Data Communications vs. Distributed Computing

1
Data Communications vs. Distributed Computing

• Dr. Craig Partridge
• Chief Scientist, BBN Technologies
• Chair, ACM SIGCOMM

2
A Quick History

• In the 1980s, the data comm community largely
  stopped leading in network application
  development
• Overwhelmed by lower layer research problems
• Other communities stepped in
• OS and distributed systems
• Supercomputing and physics

3
An unfortunate side effect

• The two fields most expert in networking dont
  talk as much as they should
• Indeed, I was invited to talk here because it was
  considered nice to have a networking
  perspective...

4
Whats new in networking

• So what have those networking guys been up to for
  the past ten years or so???
• One persons perspective
• Ive tried to focus on fun topics
• So nothing on TCP performance
• Most problems are configurational

5
Self Similarity

• Trouble with queueing theory
• By late 1980s, clear that classic models didnt
  work for data traffic
• Off by factors of 10 or 100 in queue size
  estimates
• Enter Leland, Taqqu, Willinger Wilson (93)
• Data traffic is self-similar (fractal)

6
Self Similarity Example
7
More Self Similarity

• Self-similarity means traffic smooths very slowly
• traffic at 100s sample units very similar to
  traffic at 0.01 second samples
• High peak to mean ratios

8
Self Similarity in practice

• Since 1993, weve been working to reduce self
  similarity to practice
• Confirming it exists on various types of networks
• Creating generator functions for modeling
• Understanding why it exists

9
Quality of Service

• A term whose definition is evolving
• Bandwidth guarantee?
• Loss guarantee?
• Delay guarantee?
• All three?

10
The QoS Challenge

• How to do QoS in a self-similar world?
• Old style Poisson aggregation doesnt work unless
  the network loads are very very large
• QoS Triumph
• Weighted Fair Queuing (Demers, Keshav, Shenker)
• PGPS by Parekh

11
Weighted Fair Queuing

• A delightful insight
• Transform bit-wise sharing of links into
  packetized sharing
• Work conserving!
• Nicely enough, all other work conserving schemes
  have been shown to be variants of WFQ

12
Fair Queuing Diagram
13
WFQ Diagram
14
PGPS

• Packetized General Processor Sharing
• Work by Parekh
• If traffic conforms to a (general) arrival model,
  we can derive the upper bound on queuing delay
• At high speeds, bound is nearly independent of
  number of queues in the path

15
What Next for QoS?

• WFQ is expensive to implement
• Though good approximations exist
• General feeling that WFQPGPS is overkill
• Something simpler should be possible
• The community is working through various
  statistical guarantees

16
High Performance

• Around 1991, the accepted wisdom was that IP was
  dead because routers couldnt go fast
• Now, widely accepted that routers can achieve
  petabit speeds

17
What Happened?

• Mostly, good engineering
• Router innards re-engineered for speed
• But also some new prefix lookup algorithms
• Luleå algorithm
• WashU algorithm

18
Ad-Hoc Networks

• A new and exciting area
• Imagine thousand or millions of wireless nodes in
  a room
• Theyre moving
• They need to discover and federate (securely)
• Managing signal/noise ratio vital for performance

19
More on Ad-Hoc Networks

• Odd desire to say were done
• Jini
• Existing ad-hoc routing protocols
• Yet the problems remain huge
• Device location hard (user interface harder)
• Density challenges existing protocols
• Clashes over spectrum

20
Robustness

• To keep the Internet robust we must
• Improve device reliability by factor of 10 every
  two years OR
• Improve our protocols to be more resilient
• Assuming something is always going up or down
• How to minimize impact
• In traffic
• In performance
• Can PODC community help here?

21
Lots of other initiatives

• Simulation
• How do you simulate something 100 times bigger
  than anything ever built?
• Measurement
• How much can you learn just from the edge of the
  network?
• Errors
• Packets damaged frequently, what to do?
• Anycast
• Nice idea, how do we make it real?

22
The Last Slide

• Theres lots of fun work in networking
• A lot has been happening
• A lot will happen
• Some of the problems are also of interest to the
  PODC community
• I look forward to talking with you about them.