1 2 These were the tests done to validate the Diffserv scripts. 3 This document will be updated continously. If you do more 4 thorough validation testing please post the details to the 5 diffserv mailing list. 6 Nevertheless, these tests should serve for basic validation. 7 8 AFCBQ, EFCBQ, EFPRIO 9 ---------------------- 10 11 generate all possible DSCPs and observe that they 12 get sent to the proper classes. In the case of AF also 13 to the correct Virtual Queues. 14 15 Edge1 16 ----- 17 generate TOS values 0x0,0x10,0xbb each with IP addresses 18 10.2.0.24 (mark 1), 10.2.0.3 (mark2) and 10.2.0.30 (mark 3) 19 and observe that they get marked as expected. 20 21 Edge2 22 ----- 23 24 -Repeat the tests in Edge1 25 -ftp with data direction from 10.2.0.2 26 *observe that the metering/policing works correctly (and the marking 27 as well). In this case the mark used will be 3 28 29 Edge31-cb-chains 30 ---------------- 31 32 -ftp with data direction from 10.2.0.2 33 34 *observe that the metering/policing works correctly (and the marking 35 as well). In this case the mark used will be 1. 36 37 Metering: The data throughput should not exceed 2*CIR1 + 2*CIR2 38 which is roughly: 5mbps 39 40 Marking: the should be a variation of marked packets: 41 AF41(TOS=0x88) AF42(0x90) AF43(0x98) and BE (0x0) 42 43 More tests required to see the interaction of several sources (other 44 than subnet 10.2.0.0/24). 45 46 Edge31-ca-u32 47 -------------- 48 49 Generate data using modified tcpblast from 10.2.0.2 (behind eth2) to the 50 discard port of 10.1.0.2 (behind eth1) 51 52 1) generate with src tos = 0x88 53 Metering: Allocated throughput should not exceed 2*CIR1 + 2*CIR2 54 approximately 5mbps 55 Marking: Should vary between 0x88,0x90,0x98 and 0x0 56 57 2) generate with src tos = 0x90 58 Metering: Allocated throughput should not exceed CIR1 + 2*CIR2 59 approximately 3.5mbps 60 Marking: Should vary between 0x90,0x98 and 0x0 61 62 3) generate with src tos = 0x98 63 Metering: Allocated throughput should not exceed CIR1 + CIR2 64 approximately 2.5mbps 65 Marking: Should vary between 0x98 and 0x0 66 67 4) generate with src tos any other than the above 68 Metering: Allocated throughput should not exceed CIR1 69 approximately 1.5mbps 70 Marking: Should be consistent at 0x0 71 72 TODO: Testing on how each color shares when all 4 types of packets 73 are going through the edge device 74 75 Edge32-cb-u32, Edge32-cb-chains 76 ------------------------------- 77 78 -ftp with data direction from 10.2.0.2 79 80 *observe that the metering/policing works correctly (and the marking 81 as well). 82 83 Metering: 84 The data throughput should not exceed 2*CIR1 + 2*CIR2 85 + 2*PIR2 + PIR1 for u32 which is roughly: 6mbps 86 The data throughput should not exceed 2*CIR1 + 5*CIR2 87 for chains which is roughly: 6mbps 88 89 Marking: the should be a variation of marked packets: 90 AF41(TOS=0x88) AF42(0x90) AF43(0x98) and BE (0x0) 91 92 TODO: 93 -More tests required to see the interaction of several sources (other 94 than subnet 10.2.0.0/24). 95 -More tests needed to capture stats on how many times the CIR was exceeded 96 but the data was not remarked etc. 97 98 Edge32-ca-u32 99 -------------- 100 101 Generate data using modified tcpblast from 10.2.0.2 (behind eth2) to the 102 discard port of 10.1.0.2 (behind eth1) 103 104 1) generate with src tos = 0x88 105 Metering: Allocated throughput should not exceed 2*CIR1 + 2*CIR2 106 +PIR1 -- approximately 4mbps 107 Marking: Should vary between 0x88,0x90,0x98 and 0x0 108 109 2) generate with src tos = 0x90 110 Metering: Allocated throughput should not exceed CIR1 + 2*CIR2 111 + 2* PIR2 approximately 3mbps 112 Marking: Should vary between 0x90,0x98 and 0x0 113 114 3) generate with src tos = 0x98 115 Metering: Allocated throughput should not exceed PIR1+ CIR1 + CIR2 116 approximately 2.5mbps 117 Marking: Should vary between 0x98 and 0x0 118 119 4) generate with src tos any other than the above 120 Metering: Allocated throughput should not exceed CIR1 121 approximately 1mbps 122 Marking: Should be consistent at 0x0 123 124 TODO: Testing on how each color shares when all 4 types of packets 125 are going through the edge device 126
