Capacity Planning for the Internet
Download
Report
Transcript Capacity Planning for the Internet
Capacity Planning for
Internet Service
Networks
Geoff Huston
NTW Track4
Issues
TCP/IP Protocol Behavior
Issues
Usage Profile
Capacity Guidelines
Growth Levels
Planning Issues
Understand the domain of operation
technical issues
market issues
competitive issues
regulatory issues
TCP/IP Protocol Issues
TCP/IP is NOT a flow damped protocol
end to end flow management
sliding window protocol
adaptive flow rate designed to probe and use
max available end to end bandwidth
only limited by end system buffering size
bandwidth x delay
system buffers are getting larger as OS
vendors come to understand the problem
TCP/IP Protocol Issues
TCP/IP Data Flow Rate Adaptation
Steady State Available Bandwidth
Data Flow Rate
Rate overflow loading
into network queues
Time
TCP/IP Protocol Issues
No network-based flow control mechanism
Network-based packet loss signals end systems
to collapse transmission window size
Varying window size allows adaptive flow
metrics to adapt to changing maximum available
capacity
Sustained insufficient capacity leads to
congestion induced collapse of data throughput
TCP/IP Protocol Issues
Many simultaneous TCP sessions interact
with non-predictive non-uniform load
(ftp://thumper.bellcore.com/pub/dvw/sigcom93.ps.Z)
Peaks start to synchronize with each other
Buffering evens out individual flows, but
buffers themselves behave with
fluctuating load
Buffering adds latency
TCP/IP Protocol Issues
66%
Congestive Collapse The slide to misery
and packet loss
33%
Data Throughput
TCP/IP efficiency under congestion load
33%
Traffic Level
66%
TCP/IP Protocol Issues
TCP vs UDP
UDP-based applications
Internet Phone, Video, Workgroup
UDP Issues
no flow control mechanism
sustained use forces precedence over TCP
flows
increasing use of flow bandwidth negotiated
protocols for these applications (RSVP)
TCP/IP Protocol Issues
Damping network capacity is not a
demand management tool
Network capacity must be available to
meet peak demand levels without
congestion loss
Usage Profile
Two major Internet use profiles:
Business use profile
peak at 1500 - 1600
plateau 1000 - 1730
Residential dial profile
peak at 2030 - 2330
plateau 1900 - 2400
Usage Profile
Distance profiles
12% Local
18% Domestic Trunk
70% International
Traffic mix due to:
Distance invisible applications without user
control
Distance independent user tariff
Capacity Guidelines
Link Utilisation
Average weekly traffic level set to 50% of
available bandwidth.
Core network capacity should be
dimensioned according to aggregate
access bandwidth
Link Usage Profile - optimal
peak loading less than 10% time
greater than 50% loading for 50%
time
traffic bursting visible
Link Usage Profile - overloaded
90% peak loading for 45% time
60% peak loading for 60% time
no burst profile at peak loads
imbalanced traffic (import based)
Link Usage Profile - saturated
visible plateau traffic load signature
small load increases cause widening
plateau
Overall Growth Levels
Two growth pressures:
serviced population
the changing Internet service model
more network-capable applications
using more bandwidth
Overall Growth Levels
For a constant service model the growth
curve will exhibit demand saturation
500
400
300
200
100
0
1
11
21
31
41
51
61
Overall Growth Levels
For a changing service model the saturation point will
move
More intense network use by increasingly sophisticated
applications
1200
1000
Technology shift
800
600
400
200
0
1
11
21
31
41
51
61
How to plan
Generate a market demand model
forecast the number of services in operation
existing services
growth rate
market capture level (competitive position)
forecast the average demand per service
dial access, leased line
web, ftp, usenet
caching trends
new Internet services
How to plan
demand models are typically very uncertain
indicators
high level of uncertainty of externalities
highly dynamic competitive position
poorly understood (and changing) service
demand model
How to plan
Forward extrapolation
assume existing traffic follows a general
growth model
forward extrapolat the growth model
Good for short term planning (12 months)
Cannot factor
latent demand
market price sensivity
4/27/97
4/17/97
4/7/97
3/28/97
3/18/97
3/8/97
2/26/97
2/16/97
2/6/97
1/27/97
1/17/97
1/7/97
12/28/96
12/18/96
12/8/96
11/28/96
11/18/96
11/8/96
10/29/96
10/19/96
10/9/96
9/29/96
9/19/96
9/9/96
8/30/96
8/20/96
8/10/96
7/31/96
7/21/96
7/11/96
7/1/96
Trend forecasting
historical usage vs capacity data
120000
100000
80000
60000
Total Capacity
Daily Traffic IN
Daily Traffic OUT
40000
20000
0
23/8/98
27/7/98
30/6/98
3/6/98
7/5/98
10/4/98
14/3/98
15/2/98
19/1/98
23/12/97
26/11/97
30/10/97
3/10/97
6/9/97
10/8/97
14/7/97
17/6/97
21/5/97
24/4/97
28/3/97
1/3/97
2/2/97
6/1/97
10/12/96
13/11/96
17/10/96
20/9/96
24/8/96
28/7/96
1/7/96
Growth Trends
600000
500000
400000
in
out
Trend - high
Trend - mid
300000
Trend - Low
97/98 bandwidth
200000
100000
0
Planning
undertake demand and trend forecast
models
constantly review the model against
generated usage data
recognise that the larger the capacity you
need the longer the lead time to purchase it
recognise that the bigger the purchase the
greater the requirement for capital