Download Broadband Access Systems, Inc. Cuda 12000 IP

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Power over Ethernet wikipedia , lookup

Net neutrality wikipedia , lookup

Distributed firewall wikipedia , lookup

National Broadband Plan (United States) wikipedia , lookup

TV Everywhere wikipedia , lookup

RapidIO wikipedia , lookup

Wake-on-LAN wikipedia , lookup

Net neutrality law wikipedia , lookup

IEEE 802.11 wikipedia , lookup

Asynchronous Transfer Mode wikipedia , lookup

Network tap wikipedia , lookup

Piggybacking (Internet access) wikipedia , lookup

Cracking of wireless networks wikipedia , lookup

UniPro protocol stack wikipedia , lookup

Passive optical network wikipedia , lookup

Net bias wikipedia , lookup

Throughput wikipedia , lookup

Deep packet inspection wikipedia , lookup

IEEE 1355 wikipedia , lookup

Transcript
T
H
E
TO L LY
G R O U P
No. 200212
June 2000
Broadband Access Systems, Inc.
Test
Summary
Cuda 12000 IP Access Switch
Cable Modem System Performance Evaluation
Test results show that the Cuda 12000 IP
Access Switch can forward traffic at more than
96% of the theoretical maximum for the following configurations: maximum packet forwarding rates with 64-byte packets (minimum
packet size); and maximum packet forwarding
rates using 108-byte packets (size of packetized voice solution traffic). When testing sim-
© 2000 The Tolly Group
Achieves near wire-speed small packet-processing
performance of 64- and 108-byte packets
Delivers near wire-speed simulated Internet traffic
performance
Demonstrates scalable support for hundreds of simultaneous
connections per DOCSIS module
Cuda 12000 CMTS Bidirectional Packet Processing
373 Cable Data Modems
100
99.6%
99.7%
96.5%
97.2%
98.5%
97.2%
50
97.9%
75
97.2%
B
roadband Access Systems, Inc. commissioned The Tolly Group to evaluate its
Cuda 12000 IP Access Switch, a Cable Modem
Termination System (CMTS) equipped with a
DOCSIS module (Hybrid Fiber Coax interface
to the CDM network), in conjunction with several hundred CDMs from Toshiba America
Information Systems, Inc. Tests were conducted for zero-loss, steady-state, asymmetric
throughput with full-duplex traffic using 64
bytes for maximum packet processing, 108
bytes for packet telephony and 64 bytes
upstream/1,518 bytes downstream and 200
bytes upstream and 1,518 bytes downstream to
test common IP traffic. Furthermore, engineers
tested 108- and 200-byte packets to test the
capability of the Cuda 12000 IP Access Switch
when forwarding packet sizes commonly
found in packetized voice solutions. Testing
was performed in May 2000.
Test Highlights
Percentage of
theoretical maximum
Premise: Cable System Operators (CSOs),
Local Cable Operators (LCOs), Internet
Service Providers (ISPs) and competitive local
exchange carriers (CLECs), who wish to
deploy Cable Data Modem (CDM) networks
require objective, accurate assessments of
system performance in “real-world” configurations. While single-modem performance
results can be instructive, service providers
really need to understand how the entire network will respond when traffic from dozens, or
even hundreds, of simultaneous users is active
across a CDM infrastructure. Such information
allows network managers to scale or provision
the network properly in response to changes in
the number of users, expected-traffic profile
and acceptable levels of oversubscription.
25
0
64 bytes
108 bytes
64 bytes,
1,518 bytes
200 bytes,
1,518 bytes
Packet size, bytes (incl. CRC)
Upstream
Source: The Tolly Group, June 2000
ulated Internet transactions of 200 bytes
upstream requests and 1,518-byte responses,
the Cuda 12000 IP Access Switch yielded similar performance results.
Results
Cable data modem networks differ from traditional, leased-line networks and Local Area
Networks, in that the available upstream and
Downstream
Figure 1
downstream bandwidths are asymmetrical.
While a maximum data rate of 37 Mbit/s is
possible in the downstream using 256 quadrature amplitude modulation (QAM), this data
rate is not possible in the upstream because of
natural signal impairments that accumulate in
the cable plant. For this reason, a less dense
modulation rate (16 QAM or 10 Mbit/s) is generally used in the upstream. Furthermore,
while upstream bandwidth is generally
Page 1
Broadband Access Systems Cuda 12000 IP Access Switch
shared only within a single node of a
cable data modem network, downstream bandwidth is usually shared
among several nodes connected to
the CMTS.
For this set of tests, results show the
outcome of testing asymmetrical
upstream and downstream traffic, with
upstream defined as traffic to the CMTS
and downstream defined as traffic
from the CMTS. The Cuda 12000 IP
Access Switch, with a 1 x 4 (one down,
one up) DOCSIS module, was set to
256 QAM downstream modulation and
16 QAM upstream modulation.
Upstream and
Downstream
Throughput of 64-Byte
Packets
Results show that the Cuda 12000 IP
Access Switch could forward traffic
consisting of 64-byte packets upstream
at 97% of the theoretical maximum and
downstream at 97% of the theoretical
maximum. See figure 1.
When thought of in terms of Mbit/s,
the upstream traffic of 64-byte packets
resulted in an aggregate system
throughput of 23.9 Mbit/s while the
downstream traffic of 64-byte packets
resulted in an aggregate system throughput of 33.6 Mbit/s. See figure 2.
Upstream and
Downstream
Throughput of 108Byte Packets
The Tolly Group also verified that the
Cuda 12000 IP Access Switch could
forward traffic consisting of 108-byte
packets (a size that closely matches that
of some packetized voice solutions)
upstream at 97% of the theoretical maximum and downstream at 98% of the
theoretical maximum. See figure 1.
The upstream traffic of 108-byte packets resulted in an aggregate system
throughput of 26.9 Mbit/s while the
downstream traffic of 108-byte packets
resulted in an aggregate system
throughput of 34.8 Mbit/s. See figure 2.
© 2000 The Tolly Group
Cuda 12000 CMTS Bidirectional Packet Processing
373 Cable Data Modems
34.8
33.6
35
Average system
throughput (Mbit/s)
The Tolly Group
30
26.9
23.9
25
20
15
10
5
0
64 bytes
108 bytes
Packet size, bytes (incl. CRC)
Upstream
Source: The Tolly Group, June 2000
Upstream and
Downstream
Throughput of 64- and
1,518-byte Packets
When faced with tests of 64-byte packet
traffic upstream, and 1,518-byte packet
traffic downstream, the Cuda 12000 IP
Access Switch delivered 97% of the
theoretical maximum for upstream traffic
and 96% of the theoretical maximum for
downstream traffic. See figure 1.
Upstream and
Downstream
Throughput of 200and 1,518-byte Packets
In a final testing scenario, the Cuda
12000 IP Access Switch showed that it
can forward upstream traffic using 200byte packets at 99% of the theoretical
maximum and that it could forward
downstream traffic using 1,518-byte
packets at 99% of the theoretical maximum. See figure 1.
Analysis
The test results suggest that service
providers that deploy the Cuda 12000 IP
Access Switch in conjunction with
Toshiba DAZ8801F modems can provision the network to support hundreds of
Downstream
Figure 2
simultaneous users performing highvolume transactions concurrently.
The tests demonstrate that such a
network will support near wire-speed
performance of heavy interactive traffic
(based upon the results of tests with 64byte upstream and downstream packets),
heavy packetized voice loads (based
upon the results of tests with 108-byte
upstream and downstream packets) and
heavy Internet access (based upon the
results with 64- and 200-byte packets
upstream and 1,518-byte packets
downstream).
Although the configuration of 200-byte
packets upstream and 1,518-byte packets
downstream exhibited loss above the
0.1% target of this evaluation, the 0.21%
loss ratio still represents less than a
quarter of the loss ratio deemed acceptable by Cable Television Laboratories,
Inc. (CableLabs ®), namely 1.0%. Moreover, such a loss ratio is not likely to have
a major impact on application performance since it represents an approximate
average loss of only one packet in five
hundred. Customarily The Tolly Group
tests LAN switches to .001% loss; however, this is far too stringent for cable
data modem technology. A 1.0% loss can
have a significant impact on network utilization in congested networks. For this
test, The Tolly Group conducted a more
stringent set of tests to more closely
approximate the steady-state performance
Page 2
The Tolly Group
Broadband Access Systems Cuda 12000 IP Access Switch
of congested networks with a high number of active cable modems.
Test Configuration
and Methodology
The system under test, a Broadband
Access Systems Cuda 12000 IP Access
Switch, a Cable Modem Termination
System [CMTS], version 1.0.0
cmts_cw14 9 2000_05_11_1423, connected to seven Hybrid Fiber Coax (HFC)
Cables with fiber/coax converters. Four
cables served as carrier networks that carried upstream traffic. Each carrier connected to its own diplexer, a three-port
frequency-dependent device that is used
as a separator or a combiner of signals
that fanned out each cable so it could connect to numerous Data Over Cable
Service Interface Specification (DOCSIS)
cable modems. A total of 512 modems
were configured in this test bed. Each of
four simulated networks connected to 81,
84 and two sets of 104 Toshiba American,
Inc. DAZ8801F cable modems. All were
DOCSIS 1.0 certified, hardware version
5.21 and software version 1.6.016. In
addition, 139 idle modems were connected to the networks to further simulate a
real-world environment.
Each active cable modem connected to a
half-duplex Ethernet link that connected
to a Wavetek Wandel Goltermann
DominoFastEthernet DA-350 DominoCore version 2.5 10Base-TX hardwarebased network analyzer to monitor traffic. The Domino ran on one of two of the
following during different testing sessions: a 500-MHz Pentium III Quantex
T-1401 with 160 Mbytes of RAM with a
fixed-disk size of 12 Gbytes running
Nortel Networks NetGear FA-510
10/100 Mbit/s with a PCMCIA bus; and
a Microsoft Corp. Windows 98 second
edition; or a 133-MHz Pentium Compaq
Armada 1590DMT with 32 Mbytes of
RAM and a fixed-disk space of 2.0
Gbytes, also running Microsoft Windows
98 Second Edition.
A second Wavetek Wandel Goltermann
DominoFastEthernet DA-350 DominoCore version 2.5 100Base-TX hardware-based network analyzer was connected to the network to monitor traffic.
It was running on a 300-MHz Pentium
II Dell Computer Corp. Latitude CPi
300XT with 64 Mbytes of RAM, an
OCI bus card and 4.2 Gbytes of fixeddisk space. This console was also run-
© 2000 The Tolly Group
ning a 3Com Corp. 3Com Megahertz
10/100 LAN CardBus Ethernet PC
card and Microsoft Windows NT
Workstation 4.0 SP5.
The network monitors connected to one
of two Netcom Systems, a division of
Spirent Communications Inc., SmartBits SMB-2000 Advanced Multiport
Network Performance Tester/Analyzer
/Simulators, populated with ML-7710
SmartCard interfaces, with a SmartCableModem DOCSIS compliant test
beta version 1.12, 20-port network
traffic simulator. Both SmartBits consoles ran on two separate 500-MHz
Pentium III Dell OptiPlex GX110 with
128 Mbytes of RAM. All traffic generated from the SmartBits systems to the
Cuda 12000 IP Access Switch for
testing. See figure 3.
The Cuda 12000 IP Access Switch was
configured for a single downstream
carrier of 507 MHz and four upstream
carriers of 20 MHz, 21.6 MHz, 23.2
MHz and 24.8 MHz.
Engineers tested the Cuda 12000 IP
Access Switch for zero-loss throughput.
The test was made asymmetric by
including four upstream carriers and one
downstream carrier so that the downstream traffic stream was four times the
volume of the upstream traffic streams.
Engineers conducted the following tests:
64-byte packets upstream and downstream; 108-byte packets upstream and
downstream; 64-byte packets upstream
and 1,518-byte packets downstream; and
finally, 200-byte packets upstream and
1,518-byte packets downstream.
Engineers executed three iterations for
each test and recorded data throughput
as reported by SmartBits. Note: The
Toshiba modems periodically transmitted
duplicate copies of a given frame.
Therefore, The Tolly Group calculated
the number of packets forwarded by the
system based upon the number of frames
received in sequence, not the total
number received, as reported by
SmartCableModem Test. This methodology therefore excluded from its count
any frames that were forwarded uniquely
by the system error-free but out of
sequence. It also reduced, but did not
eliminate, the possibility that duplicate
frames would be counted.
Broadband
Access
Systems, Inc.
Cuda 12000
IP Access
Switch
Cable Modem
System Performance Evaluation
Broadband Access Systems, Inc.
Cuda 12000 IP Access Switch
Product Specifications*
Cuda 12000TM IP Access Switch
14-slot chassis (9RU)
12 slots for application modules
MeshFlowTM high-speed interconnect
200+ Gbit/s peak switching throughput
Dual -48v DC connectors
Access Modules
DOCSIS module, one downstream, four
upstream
Programmable Layer 3 switch engine
DOCSIS 1.0 qualified, 1.1 ready
Integrated up-converter
SNMPv2 managed
Downstream modulation QAM64/256
Downstream rate 27 or 38 Mbit/s
Upstream demodulation QPSK/QAM16
Upstream Rate 320 Kbit/s to 10 Mbit/s
Egress Module Options
Octal 10/100Base-TX module
Programmable Layer 3 switch engine
Routing Protocols: RIPv1/2, OSPFv2
Single 1000Base-SX/LX module
Single OC3/12 POS module
Management
Chassis controller module
Console: SNMP, HTTP, ICMP, telnet
Optional CudaViewTM Manager
Optional Cuda Provisioning ManagerTM
Standards
DOCSIS 1.0
BellCore GR-1089, GR-63 CORE
IETF IPv4
IEEE 802.3
For more information contact:
Broadband Access Systems, Inc.
8 Technology Drive
Westborough, MA 01581
Phone: (508) 366-8833
Fax: (508) 836-2677
URL: http://www.basystems.com
*Vendor-supplied info not verified by The
Tolly Group
Page 3
The Tolly Group
Broadband Access Systems Cuda 12000 IP Access Switch
Test Bed
SmartBits SMB-2000
75 ohm RF Taps
SD
23
23
2000
PO W ER
75 ohm
LI NK
1
2
1 52 27
A T-9 62 2s
4
A T-96 2 2s
Per for m ance A naly sis S ys te m
9
10
11
14
15
16
17
18
19
20
TX
TX
TX
TX
TX
TX
TX
TX
TX
TX
TX
INI T
INIT
I NIT
I NI T
INI T
INI T
INIT
I NI T
INI T
INI T
INI T
INI T
TX
TRIG
TRIG
T RIG
T RIG
TRIG
TRI G
T RIG
T RIG
TRI G
TRI G
TRIG
T RIG
TRI G
TRI G
TRI G
ALRM
CL S
CLS
CLS
C LS
CLS
CLS
CLS
C LS
CLS
CL S
CLS
C LS
CLS
CLS
CLS
CLS
RX
LOS
TRI G
RX
ERR
RX
ERR
RX
E RR
RX
E RR
RX
ERR
RX
ERR
RX
E RR
RX
E RR
RX
ERR
RX
ERR
RX
ERR
RX
E RR
RX
ERR
RX
ER R
RX
E RR
RX
E RR
TX
DP LX
100M
1300nm
LINK
DP LX
100M
LI NK
DP LX
LI NK
LINK
D PLX
LINK
DPLX
L INK
DP LX
100M
LI NK
DP LX
LINK
D PLX
LINK
DPLX
100M
LINK
DP LX
LI NK
DP LX
LINK
D PLX
LI NK
DPLX
T IG
R
LINK
DP LX
LINK
DPL X
DPLX
100M
100M
100M
100M
100M
100M
100M
100M
100M
100M
100M
PA US E
PA US E
P AUS E
PA USE
PA US E
P AUS E
P AUS E
P AUS E
PA US E
P AUS E
P AUS E
P AUS E
PA US E
P AUS E
P AUS E
P AU SE
N ET C O M
SYST E M S
N ET C O M
S YST E M S
N ET C O M
S YS T EM S
N ET C OM
SYST EM S
N E TC O M
SYST E M S
NET C O M
S YST E M S
NET C O M
SYS T EM S
N ET C O M
SYS T EM S
N E T CO M
SYST E M S
NET CO M
SY ST E M S
NET C O M
SYS T EM S
N ET C O M
SYS T EM S
N E T CO M
SYST EM S
NE T CO M
SY ST EM S
NET C O M
SYS T EM S
NET C O M
SYS T EM S
100M
Clas s 1
LA SE R
P RO DUCT
RX
NET C O M
SY ST EM S
RX
S ing le M od e
N ET C O M
S YST E M S
SD
23
2000
PO W ER
75 ohm
LI N K
1
2
A T-9 62 2 s
3
4
A T-9 62 2s
TX
5
15227
I NI T
TX
TX
INI T
INIT
TRI G
TRIG
TX
NET C O M
SYST EM S
Half-duplex Ethernet
links, 1 per modem
Po rt
Stre a m
La ye r
9
TX
Pe rfo rm anc e A naly sis Sys te m
10
11
12
TX
13
S Y S T E MS
14
15
16
TX
17
18
19
20
TRI G
T RI G
TRI G
CLS
CLS
CLS
CLS
CLS
CLS
RX
ER R
RX
E RR
RX
ERR
RX
ER R
RX
E RR
RX
E RR
RX
E RR
LINK
DPLX
LINK
L INK
DP L X
LIN K
D PLX
DPLX
LI NK
DPLX
LINK
DP LX
LINK
DPLX
LIN K
DPLX
LI NK
DP LX
LI NK
DP LX
INI T
TX
TRI G
CL S
RX
ERR
DPLX
INI T
TX
TRIG
CLS
RX
ERR
LINK
I NI T
TX
TRI G
CLS
RX
E RR
LI NK
IN IT
TX
INI T
TRI G
C LS
RX
ERR
D PLX
I NI T
TX
INI T
TRI G
CLS
RX
ERR
DP LX
INI T
TX
T RIG
CLS
RX
ERR
LINK
I NI T
TX
TRI G
CLS
RX
E RR
DP LX
INIT
TX
TRI G
C LS
RX
ERR
TX
INI T
TX
INI T
CLS
1300nm
Clas s 1
LA S ER
P RODUCT
RX
Sin gle M o de
Multi
8
TX
TRIG
CL S
RX
ERR
TRI G
100M
1300nm
Clas s 1
LA SE R
P RODUCT
7
TX
I NI T
T RIG
ALR M
RX
LOS
LINK
Toshiba DOCSIS
Cable Modems
6
ML-77 1 0 ML-7 71 0 M L-7 71 0 M L-7 71 0 ML-77 1 0 ML-7 71 0 ML-7 71 0 M L-7 7 10 ML-7 71 0 ML-7 71 0 ML-7 71 0 ML-7 7 10 ML- 77 10 ML-7 71 0 ML-7 71 0 M L-7 7 10
TX
IN IT
TRI G
A LRM
RX
LOS
2
S Y S T E M S
13
TX
I NI T
Si n gle M od e
Diplexer
12
TX
I NIT
Clas s 1
LAS E R
P RODU CT
75 ohm RF Taps
Por t
S tre am
La ye r
Multi
8
TX
INIT
1300nm
Half-duplex Ethernet
links, 1 per modem
7
TX
INI T
A LRM
LINK
Toshiba DOCSIS
Cable Modems
6
TX
I NI T
TRIG
RX
LOS
23
5
M L-77 10 ML- 77 1 0 ML -7 71 0 M L-7 71 0 M L-77 10 ML-77 1 0 ML-7 71 0 M L-7 71 0 M L-77 1 0 ML-7 71 0 ML-7 71 0 M L-7 71 0 ML-7 7 10 ML-7 71 0 ML-7 71 0 ML-7 71 0
TX
INI T
15227
Active Hybrid Fiber Coax (HFC) Cable Modem
Network with Fiber/Coax Converters
Diplexer
3
TX
1 52 27
1
LINK
DPL X
IN IT
T IG
R
LIN K
DPLX
100M
100M
100M
100M
100M
100M
100 M
100M
100M
100M
10 0M
100M
100M
100M
100M
PA US E
P AUS E
P AUS E
PA US E
P AUS E
P AUS E
P AU SE
P A US E
P AUS E
P AUS E
P AU SE
P A US E
P A US E
P AUS E
P AU SE
P A USE
N ET CO M
SYST E M S
NET C O M
S YS T EM S
N ET C O M
SYS T EM S
N E TC O M
SYST EM S
NE T CO M
SYST E M S
NET C O M
S YS T EM S
NET C O M
SYS T EM S
N E T CO M
SYS T EM S
NE T CO M
SY ST E M S
NET CO M
SY ST EM S
NET C O M
SYS T EM S
N E T CO M
SYS T EM S
NE T CO M
SY ST EM S
NE T CO M
SY ST EM S
NET C O M
SYS T EM S
N ET CO M
SYS T EM S
RX
S ing l e M od e
N E T CO M
S YST E M S
23
= 23 dB Tap
3
Diplexer
4
Diplexer
Broadband Access Systems
Cuda 12000 IP Access Switch
(CMTS)
Wavetek Wandel
Goltermann
DominoFastEthernet
Internetworking
Analyzer DA-350
Carrier = 507 MHz
Carrier 1 = 20.0 MHz
Carrier 3 = 23.2 MHz
Carrier 2 = 21.6 MHz
Carrier 4 = 24.8 MHz
Wavetek Wandel
Goltermann
DominoFastEthernet
Internetworking
Analyzer DA-350
Source: The Tolly Group, June 2000
1
= 36 km upstream,
36 km downstream,
81 modems active
2
= 15 km upstream,
9 km downstream,
84 modems active
3
= 18 km upstream,
18 km downstream,
104 modems active
4
= 15 km upstream,
9 km downstream,
104 modems active
2 Half-duplex Fast
Ethernet links
Figure 3
The Tolly Group gratefully acknowledges the providers
of test equipment used in this project.
Vendor
Netcom Systems, Inc.
Netcom Systems, Inc.
Wavetek Wandel Goltermann
Product
SmartBits SMB-2000
SmartCableModem
DominoFastEthernet
Since its inception, The Tolly
Group has produced highquality tests that meet three
overarching criteria: All tests
are objective, fully documented
and repeatable.
We endeavor to provide complete disclosure of information
concerning individual product
tests, and multiparty competitive product evaluations.
As an independent organization, The Tolly Group does not
accept retainer contracts from vendors, nor does it endorse
products or suppliers. This open and honest environment
assures vendors they are treated fairly, and with the
necessary care to guarantee all parties that the results of
these tests are accurate and valid. The Tolly Group has
codified this into the Fair Testing Charter, which may be
viewed at http://www.tolly.com.
Web address
http://www.netcomsystems.com
http://www.netcomsystems.com
http://www.wwg.com
Project Profile
Sponsor: Broadband Access Systems, Inc.
Document number: 200212
Product class: Cable Data Modem switching
Products under test:
Cuda 12000 IP Access Switch
Testing window: May 2000
Software versions tested:
V. 1.0.0.cmts_cw14 9 2000_05_11_1423
Software status:
Generally available
For more information on this document, or other
services offered by The Tolly Group, visit our World
Wide Web site at http://www.tolly.com, send E-mail
to [email protected], call (800) 933-1699 or
(732) 528-3300.
Internetworking technology is an area of rapid growth and constant change. The Tolly Group conducts engineering-caliber testing in
an effort to provide the internetworking industry with valuable information on current products and technology. While great care is
taken to assure utmost accuracy, mistakes can occur. In no event shall The Tolly Group be liable for damages of any kind including
direct, indirect, special, incidental, and consequential damages which may result from the use of information contained in this document. All trademarks are the property of their respective owners.
The Tolly Group doc. 200212 rev. clk 27 Jun 00
© 2000 The Tolly Group
Page 4