Scheduling

1
Scheduling

• Source S. Keshav, An engineering approach to
  computer networking.

2
Outline

• What is scheduling
• Requirements of a scheduling discipline
• Fundamental choices
• Scheduling best effort connections
• Scheduling guaranteed-service connections
• Packet drop strategies

3
Scheduling

• Sharing always results in contention
• A scheduling discipline resolves contention
• whos next?
• Key to sharing resources
• Example
• on application level, consider queries awaiting
  web server
• on transport level, consider packets waiting to
  be put on a link at output queues of a switch
  (will be studied in the rest of the chapter)

4
What can scheduling disciplines do?

• Give different users different qualities of
  service
• Scheduling disciplines can allocate
• bandwidth
• delay
• loss
• They also determine how fair the network is

5
Outline

• What is scheduling
• Requirements of a scheduling discipline
• Fundamental choices
• Scheduling best effort connections
• Scheduling guaranteed-service connections
• Packet drop strategies

6
Requirements

• An ideal scheduling discipline
• is easy to implement
• is fair
• What does it means for packets scheduling at the
  output of switches?

7
Requirements 1. Ease of implementation

• Scheduling discipline has to make a decision once
  every few microseconds!
• Should be implementable
• for hardware VLSI critical constraint is size of
  the scheduling state (pointer to packet queues,
  memory about services already received by
  connections, )
• Work per packet should scale less than linearly
  with number of active connections

8
Requirements 2. Fairness

• Scheduling discipline must allocate a resource
  fairly
• Intuitively
• each connection gets no more than what it wants
• the excess, if any, is equally (weighted) shared
• Fairness also provides protection

Transfer half of excess
Unsatisfied demand
A
B
A
B
C
C
9
Outline

• What is scheduling
• Requirements of a scheduling discipline
• Fundamental choices
• Scheduling best effort connections
• Scheduling guaranteed-service connections
• Packet drop strategies

10
Fundamental choices

• 1. Degree of aggregation ( number of classes of
  traffic)
• Within a class of traffic
• 2. Work-conserving vs. non-work-conserving
• If more than a single class of traffic
• 3. Number of priority levels
• 4. Service order

11
Choices 1. Degree of aggregation

• More traffic aggregation
• less state
• cheaper
• smaller VLSI
• less to advertise
• BUT less individualization, no protection within
  a class
• Single class in current Internet!

12
Choices 2. Work conserving vs.
non-work-conserving

• Work conserving discipline is never idle when
  packets await service
• Non-work conserving key conceptual idea delay
  packet till eligible
• Why bother with non-work conserving?
• Reduces delay-jitter gt fewer buffers in network

FCFS
13
Choices 3. Priority

• Packet is served from a given priority level only
  if no packets exist at higher levels (multilevel
  priority with exhaustive service)
• Highest level gets lowest delay
• Watch out for starvation!
• Usually priority levels correspond to delay
  classes
• Low bandwidth urgent messages
• Realtime
• Non-realtime

Priority
14
Choices 4. Service order of classes with same
priority

• In each aggregation class, packets are served in
  order of arrival (First Come First Served)
• If one single class,
• bandwidth hogs win (no protection)
• If several classes, scheduling scheme decides
  which class to serve next
• protection

15
Outline

• What is scheduling
• Requirements of a scheduling discipline
• Fundamental choices
• Scheduling best effort connections
• Scheduling guaranteed-service connections
• Packet drop strategies

16
Scheduling single class best-effort connection
(Internet)

• Single class -gt no priority and no real
  scheduling packet are served to the output link
  in their order of arrival
• Implementation Single FIFO (First In First Out)
  queue
• Very simple
• Unfair (Fairness will require cooperation between
  users. So, Internet will rely on trust and
  respect)

17
Scheduling best-effort connections More than one
class in a priority level

• Main requirement is fairness
• Achievable using Generalized processor sharing
  (GPS)
• Visit each non-empty queue in turn
• Serve infinitesimal from each (impossible in
  practice!)
• Why is this fair?
• How can we give weights to connections?

18
Simple GPS emulation Weighted round robin

• Serve a packet from each non-empty queue in turn
• Unfair if packets are of different length or
  weights are not equal
• Different weights, fixed packet size
• serve more than one packet per visit, after
  normalizing to obtain integer weights
• Different weights, variable size packets
• normalize weights by mean packet size
• e.g. weights 0.5, 0.75, 1.0, mean packet sizes
  50, 500, 1500
• normalize weights 0.5/50, 0.75/500, 1.0/1500
  0.01, 0.0015, 0.000666, normalize again 60,
  9, 4
• More complex in practice (average size of packets
  is unknown, variable weights, ...)

19
Outline

• What is scheduling
• Why we need it
• Requirements of a scheduling discipline
• Fundamental choices
• Scheduling best effort connections
• Scheduling guaranteed-service connections
• Packet drop strategies

20
Scheduling guaranteed-service connections

• With best-effort connections, goal is fairness
• With guarantee-service, there is admission
  control of connections and goal is to provide
  sufficient bandwidth or delay to admitted
  connections
• We now present two scheduling discipline examples
  that provide performance guarantee

21
Example 1 Weighted Round Robin

• The highest the weight of a connection, the more
  it will be served and the higher its bandwidth
  and the smaller its delay
• But, no possibility to guarantee the delay
  without plying on bandwidth ( coupling)

22
Example 2 Rate-controlled scheduling

• A class of disciplines
• two components regulator and scheduler
• incoming packets are placed in regulator where
  they wait to become eligible ( admission
  control)
• then they are put in the scheduler
• Regulator shapes the traffic (bandwidth),
  scheduler provides performance (delay, jitter, )
  guarantees
• Decoupling between delay and bandwidth Can give
  a low-bandwidth connection a low delay without
  overbooking

23
Summary

• Two sorts of applications best effort and
  guaranteed service
• Best effort connections require fair service
• provided by GPS, which is unimplementable
• emulated by WRR and its variants
• Guaranteed service connections require
  performance guarantees
• provided by WRR, but this is not flexible enough
• may be better to use rate-controlled schedulers

24
Outline

• What is scheduling
• Why we need it
• Requirements of a scheduling discipline
• Fundamental choices
• Scheduling best effort connections
• Scheduling guaranteed-service connections
• Packet drop strategies

25
Packet dropping

• Packets that cannot be served immediately are
  buffered
• Full buffers gt packet drop strategy
• Packet losses happen almost always from
  best-effort connections (why?)
• Shouldnt drop packets unless imperative
• packet drop wastes resources (why?)

26
Early or late drop?

• Early drop gt drop as buffer is filling
• signals endpoints to reduce rate
• Late drop
• drop only when buffer is full (lots of losses in
  short time, rather unstable)

27
Drop position

• Can drop a packet from head or tail position in
  the queue
• Tail
• easy
• default approach
• Head
• harder
• lets source detect loss earlier