Network Processing Forum Benchmarking Working Group
WARRANTIES OF MERCHANTABILITY, TITLE OR FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT OF THIRD PARTY RIGHTS.
NP/CP Task Group
2
Network Processing Forum Benchmarking Working Group
Table Of Contents Benchmark Number Reporting Requirements ......................................................51 Reference Design Details ..................................................................................51.1Block Diagram ........................................................................................51.2Component List ......................................................................................51.3Mechanical dimensions ..........................................................................61.4Media, Fabric and Control interfaces ......................................................61.5Total Power Consumption ......................................................................61.6 ......................................................................................6Control Memory1.7Test Equipment ......................................................................................61.8Forwarding Table and Traffic Details ......................................................6Note: In all tests a separate graph should be included for each traffic combination tested. ...........................................................................................72 Data Plane Benchmark Tests ............................................................................82.1 Forwarding Rate....8................................................................................2.2 Throughput11............................................................................................2.3 Latency..................................41.................................................................2.3.1 IPv4 Only Latency........................................1....4................................2.3.2 IPv6 Only Latency.........................................................................1.6..2.3.3 IPv4 Concurrent with IPv6 Latency................9.1.............................2.3.3 IPv4 Concurrent with IPv6 Latency.................................19.............2.4 Loss Rate...........24...................................................................................2.4 Loss Rate.24.............................................................................................2.4.1 IPv4 Only Loss Rate........2.4...............................................................2.4.2 IPv6 Only Loss Rate.......................................................................4.22.4.3 IPv4 Concurrent with IPv6 Loss Rate5.......2...................................2.5 Overload Forwarding Rate..................................62..............................2.5.1 IPv4 Only Overload Forwarding Rate..........................26................2.5.1 IPv6 Only Overload Forwarding Rate72..........................................2.5.3 IPv4 Concurrent with IPv6 Overload Forwarding Rate.....8.......22.6 System Power Consumption................9.2............................................3 Control Plane Benchmark Tests ......................................................................303.1 Forwarding Table Update Rate........30................................................3.1.1 IPv4 Forwarding Table Update Rate.........................................3..03.1.2 IPv6 Forwarding Table Update Rate...........31................................3.2 Forwarding Rate With Concurrent Forwarding Table Updates.323.2.1 IPv4 Forwarding Rate With Concurrent IPv4 Forwarding Table Updates........32.....................................................................................3.2.2 IPv6 Forwarding Rate With Concurrent IPv6 Forwarding Table Updates................................36.............................................................
NP/CP Task Group
3
Network Processing Forum Benchmarking Working Group
3.2.2 IPv4 Concurrent with IPv6 Forwarding Rate With Concurrent IPv4 and IPv6 Forwarding Table Updates..................04..........................
NP/CP Task Group
4
Network Processing Forum Benchmarking Working Group
Benchmark Number Reporting Requirements When reporting performance numbers identified as following the NPF IP Forwarding Benchmark IA, the following text must be included: [vendor name] achieved a [benchmark metric] of [number] using the NPF IP Forwarding Benchmark employing [number of NPs][NP name] where bracketed items are filled in. For example XYZ Corporation achieved a throughput score of 1 Gb/sec using the NPF IP forwarding benchmark employing 3, XYZ45 network processors. 1 Reference Design DetailsThe IP forwarding application-level benchmark was run on a reference design with 8 1/100 Mbps Ethernet media interfaces and one PCI control interface. The reference design supports IPv4 and IPv6 protocols. The details of the reference design are given below. 1.1 Block Diagram
D e v i c e U n d e r T e s t ( D U T )
P o r t 0 P o r t 1 P o r t 2 P o r t 3 P o r t 4 P o r t 5 P o r t 6
P o r t 0 P o r t 1 P o r t 2 P o r t 3 P o r t 4 P o r t 5 P o r t 6
T r a f f i c T e s t e r
Figure 1: DUT-Traffic Tester Connections
1.2 Component List •1 NPF2002 Network Processor •64 MB of 133 MHz DDR DRAM organized in 4 banks (Part# DDRnnn) •MB of 10ns SRAM (Part# SRAMnnn)16 •Mbps Ethernet ports with NPF1008 MAC device.8 10/100 •1 32-bit, 33 MHz PCI interface. •1 Fabric Interface Chip (Part# NPF-Fabricxxxx)
NP/CP Task Group
P o r t 7
P o r t 7
5
Network Processing Forum Benchmarking Working Group
1.3 Mechanical dimensions •Standard PCI card1.4 Media, Fabric and Control interfaces •Media interfaces: 8 x 10/100 Mbps Ethernet ports•Control interfaces: 32-bit, 33 MHz, PCI•Fabric interface: None1.5 Total Power Consumption •Idle State Power Consumption5 Watts• 1.6 Control Memory •IPv4 Control Memory •1 MB SRAM for routing lookup table •256 KB DRAM for next hop forwarding table •IPv6 Control Memory •3 MB SRAM for routing lookup table •256 KB DRAM for next hop forwarding table1.7 Test Equipment •Data plane tester: IXIA 1600 •Control plane tester: PC with NPF IPv4 API software connected via PCI to the reference design 1.8 Forwarding Table and Traffic Details • used to generate traffic: 2345678 Seed • of route table used: Type o IPv4: Unmodifed Telestra AS1221 o IPv6: Unmodified AS4554 •Total Number of forwarding table entries present on the DUToIPv4Across the whole system : 120,000 :Per DUT Test Port 15,000oIPv6Across the whole system : 400Per DUT Test Port : 50•Number of route table entries exercised by traffic for all packet sizes are as followso IPv4 Across the whole system 16,000 : : 2,000 est Porter DUTP T o IPv6 Across the whole system : 400Per DUT Test Port : 50•Traffic Flow descriptionoThe Traffic Generator to DUT flows are setup as followsEach DUT Port(i) is connected to Traffic Generator Port(i)
NP/CP Task Group
6
Network Processing Forum Benchmarking Working Group
Each DUT and Traffic Generator port is setup in full duplex mode, and hence and receive and transmit packets at the same timeThe Traffic and route tables are setup as follows•All Traffic send in on DUT Port 0 (by Traffic Generator Port 0) is routed to DUT Port 1 (to Traffic Generator Port 1) and vice versa•The same is true for DUT Port 2 and 3, 4 and 5, 6 and 7
Note: In all tests a separate graph should be included for each traffic combination tested.
NP/CP Task Group
7
Network Processing Forum Benchmarking Working Group
2 Data Plane Benchmark Tests 2.1 Forwarding Rate 2.1.1 IPv4 Only Forwarding Rate1200000 1000000 800000 600000 400000 200000 0 0
Theoretical maximum Base traffic Control IP options
400 800 1200 1600 Frame size (bytes)
Figure 2: Results of Forwarding Rate Benchmark Measured in pps
800
600
400 200 0 0
Theoretical maximum Base traffic Control IP options 400 800 1200 1600 Frame size (bytes)
Figure 3: Results of Forwarding Rate Benchmark Measured in Mbps Forwarding Rate with Internet Mix Traffic = 800 Mbps
NP/CP Task Group
8
Network Processing Forum Benchmarking Working Group
Theoretical maximum Base traffic Control IP options
Figure 4: Results of Forwarding Rate Benchmark Measured in pps
800
600
400 200 0 0
Theoretical maximum Base traffic Control IP options 400 800 1200 1600 Frame size (bytes)
Figure 5: Results of Forwarding Rate Benchmark Measured in Mbps Forwarding Rate with Internet Mix Traffic = 600 Mbps
NP/CP Task Group
9
Network Processing Forum Benchmarking Working Group
2.1.3 IPv4 Concurrent with IPv6 (50% IPv4, 50% IPv6) Forwarding Rate 1200000 Theoretical maximum 1000000 Base traffic 800000 Control IP options 600000 400000 200000 0 0 400 800 1200 1600 Frame size (bytes)
Figure 6: Results of Forwarding Rate Benchmark Measured in pps
800
600
400 200 0 0
Theoretical maximum Base traffic Control IP options 400 800 1200 1600 Frame size (bytes)
Figure 7: Results of Forwarding Rate Benchmark Measured in Mbps Forwarding Rate with Internet Mix Traffic = 750 Mbps Note: The min packet size for IPv4 and IPv6 differs. Hence the frame sizes used in that case for IPv4 and IPv6 will not be the same.
NP/CP Task Group
10
Network Processing Forum Benchmarking Working Group