Ethernet-to-the-Factory 1.2 Design and Implementation Guide
Characterization of the EttF Cell/Area Zone Design
Downloads: This chapterpdf (PDF - 801.0KB) The complete bookPDF (PDF - 9.79MB) | Feedback

Characterization of the EttF Cell/Area Zone Design

Table Of Contents

Characterization of the EttF Cell/Area Zone Design

STP Testing

STP Test Methodology

STP Test Topology

STP Test Scenarios

Test Suite 1—Bidirectional Traffic (Tx1 <-> Tx2)

Test Suite 2—Bidirectional Traffic (Tx3 <-> Tx4)

Test Tools

STP Test Results

Suite 1 Test Cases

Suite 2 Test Cases

Sample Trend Line for Link Failure Between Adjacent Switches

Sample Trend Line for Link Failure To Root Bridge

16-Switch Ring—STP Testing

Test Suite 1—Bidirectional Traffic from (Tx1 <-> Tx2)

Test Suite 2—Bidirectional Traffic (Tx3 <-> Tx4)

Redundant Star Topology—STP Testing

Latency/Jitter Testing

IGMP Testing

IGMP Snooping Test Methodology

IGMP Snooping Test Topology

IGMP Snooping Test Results


Characterization of the EttF Cell/Area Zone Design


All factory floor devices are connected at the cell/area zone layer of EttF 1.1. From a network design perspective, a solid spanning tree and multicast design are critical for reliability and to meet predefined service-level agreements (SLAs). As mentioned in earlier sections, Cisco recommends that Rapid Spanning Tree Protocol (RSTP 802.1w) and IGMP snooping with querier be deployed at this layer. This appendix outlines the validation methodology and the corresponding results of the testing.

STP Testing

STP Test Methodology

Eight Cisco Catalyst 2955 switches, each running Native IOS 12.1(22)EA9, plus one Catalyst 3750 running Native IOS 12.2(25)SEB4, are linked together in back-to-back fashion via 802.1q Gigabit Ethernet trunks to form a ring topology (See Figure A-1).

All nine devices are running RSTP 802.1w carrying only one VLAN (vlan 20). This topology creates a Layer 2 loop, and therefore one port must be blocked by STP. By configuring switch CZ-3750-1 with the lowest spanning tree priority, it is elected as the root bridge for this VLAN. The STP parameters on all the other switches are left at their default settings. Accordingly, switch Cell-2955-4 is now the furthest away from the root bridge in terms of path cost and must select a port to block (GigabitEthernet 0/2). The root port on devices Cell-2955-1 through Cell-2955-4 is GigabitEthernet 0/1, and the root port on devices Cell-2955-5 through Cell-2955-8 is GigabitEthernet 0/2. A traffic generator (Ixia 400 Tf) is attached to switches Cell-2955-7, Cell-2955-8, Cell-2955-4, and Cell-2955-5 via ports Tx1, Tx2, Tx3, Tx4 respectively. The traffic generator is used to measure the convergence time in various failure scenarios.

STP Test Topology

Figure A-1 shows the test topology.

Figure A-1 Test Topology

Multiple bidirectional traffic streams are configured on Ixia between Tx1 <-> Tx2 and Tx3 <-> Tx4. Each of these pairs is referred to as a traffic suite with their own test cases. Each stream for each test suite is designed to source 1, 50, 100, and 200 MAC addresses destined to 1, 50, 100, and 200 MAC addresses. Thus, twice the number of configured source MAC addresses are traversing the ring for each test case. Each packet is a mix of 500 and 64 bytes in length sent in a continuous fashion until the STP has re-converged and the network has stabilized. At steady state for Suite 1 (Tx1 <-> Tx2), the traffic flow is as follows:

1. Packet egresses Tx1 of Ixia and ingresses port fa0/1 of Cell-2955-7

2. Packet then egresses gi0/2 on Cell-2955-7 and ingresses port g0/1 on Cell-C2955-8

3. Packet then egresses fa0/2 on Cell-C2955-8 and finally ingresses Tx2 of Ixia

This flow is reversed for streams going in the opposite direction (Tx2 to Tx1).

At steady state for Suite 2 (Tx3 <-> Tx4), the traffic flow is as follows:

1. Packet egresses Tx3 of Ixia and ingresses port fa0/3 of Cell-C2955-4

2. Because STP is blocking port gi0/2, the packet egresses gi0/1 of Cell-C2955-4 and traverses the entire ring in the clockwise direction until it reaches Cell-C2955-5

3. Packet ingresses gi0/2 on Cell-C2955-5 and egresses fa0/4

4. Packet finally ingresses Tx4 of Ixia

This flow is reversed for streams going in the opposite direction (Tx4 to Tx3).

STP Test Scenarios

Two test suites are explored, each with a different traffic flow. Within each test suite, multiple failure scenarios are introduced to simulate various STP changes in the ring topology. With each failure, convergence time is measured using the following formula:

[(Tx - Rx) / packet rate] * 1000

Where:

Tx = Packets transmitted

Rx = Packets received

PPS = 10,000 pps

Test Suite 1—Bidirectional Traffic (Tx1 <-> Tx2)

At steady state, traffic is flowing bidirectionally point-to-point between Cell-C2955-7 and Cell-C2955-8 (essentially a best-case scenario). (See Figure A-2.)

Figure A-2 Test Suite 1—Bidirectional Traffic Flow

However, after simulating a failure between these two switches, the traffic must then traverse the entire ring (becoming the worst-case scenario) to reach its destination. (See Figure A-3.)

Figure A-3 Test Suite 1—Worse-Case Scenario

The eight failure scenarios in Suite 1 are as follows:

Failure 1—Software shut link between Cell-C2955-7 and Cell-C2955-8

Failure 2—Software unshut link between Cell-C2955-7 and Cell-C2955-8

Failure 3—Physically remove link between Cell-C2955-7 and Cell-C2955-8

Failure 4—Physically re-insert link between Cell-C2955-7 and Cell-C2955-8

Failure 5—Root bridge down

Failure 6—Root bridge up

Failure 7—Stack master down on CZ-C3750

Failure 8—Stack master re-established on CZ-C3750

For each failure scenario, the following is measured and verified:

Convergence time

Verify Rockwell Automation (RA) equipment functioning properly after disruption

Measure CPU and memory on cell devices that are sending and receiving Ixia traffic

Test Suite 2—Bidirectional Traffic (Tx3 <-> Tx4)

At steady state, traffic is flowing bidirectionally between Cell-C2955-4 and Cell-C2955-5 across the entire ring. (See Figure A-4.)

Figure A-4 Test Suite 2—Bidirectional Traffic Flow

Because STP is blocking gi0/2 on Cell-C2955-4, this constitutes the worse-case scenario before any failures are introduced. However, after a network failure and a subsequent STP topology change, traffic flows to its directly-connected neighbor and becomes the best-case scenario. (See Figure A-5.)

Figure A-5 Test Suite 2—Best-Case Scenario

The eight failure scenarios in Suite 2 are as follows:

Failure 1—Software shut link between Cell-C2955-1 and CZ-C3750

Failure 2—Software unshut link between Cell-C2955-1 and CZ-C3750

Failure 3—Physically remove link between Cell-C2955-1 and CZ-C3750

Failure 4—Physically re-insert link between Cell-C2955-1 and CZ-C3750

Failure 5—Root bridge down

Failure 6—Root bridge up

Failure 7—Stack master down on CZ-C3750

Failure 8—Stack master re-established on CZ-C3750

For each failure scenario, the following is measured and verified:

Convergence time

Verify RA equipment functioning properly after disruption

Measure CPU and memory on cell devices that are sending and receiving Ixia traffic

Test Tools

The following equipment is needed for performing these tests.

16 Cisco Catalyst C2955T-12 industrial switches

2 Cisco Catalyst WS-C3750G-24PS (stacked)

1 Ixia traffic generator

1 set of RA equipment

STP Test Results

Suite 1 Test Cases

Table A-1 Test Case 1—Software Shut Link between Cell-C2955-7 and Cell-C2955-8

Run #
Baseline
A to B
Baseline
B to A
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

444.92

446.60

1105.3

1107.4

1436.7

1488

2015

2017.1

2

408.744

410.49

1362.5

1364.5

1264

1266.1

1750.5

1747.5

3

319.11

320.81

1080.6

1082.6

1064

1066

1027.4

1029.4

Avg

390.9252242

392.6380775

1182.8

1184.833333

1254.9

1273.366667

1597.633333

1598


Table A-2 Test Case 2—Software Unshut Link between Cell-C2955-7 and Cell-C2955-8

Run #
Baseline
A to B
Baseline
B to A
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

229.29

232.18

790.9

802.7

1990.6

2005.4

2005.4

1486.9

2

171.02

174.62

992.3

1002.9

1993.2

2004.8

2004.8

1584.6

3

314.75

317.79

993.6

1003.5

1987.9

2005.9

2005.9

1981.7

Avg

238.35

241.53

925.60

936.37

1990.57

2005.37

2005.37

1817.73


Table A-3 Test Case 3—Physically Remove Link between Cell-C2955-7 and Cell-C2955-8

Run #
Baseline
A to B
Baseline
B to A
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

511.44

510.21

2071.7

2071.7

1738.3

1738.3

2140.6

2140.6

2

465.48

462.35

2197.5

2197.5

1970.6

1970.6

2505.1

2505.1

3

441.89

439.76

1952.6

1952.6

2040.6

2110.2

1803

1803

Avg

472.94

470.77

2073.93

2073.93

1916.5

1939.7

2149.57

2149.57


Table A-4 Test Case 4—Physically Re-insert Link between Cell-C2955-7 and Cell-C2955-8

Run #
Baseline
A to B
Baseline
B to A
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

174.017

175.616

1810.6

1822.6

1989.9

2001

1979.3

1989.8

2

474.074

473.718

1989.7

2002.3

1863

1879.1

1986.3

2003.1

3

473.340

473.079

1992.1

2004.1

1968.3

2001.1

1891.7

1916.8

Avg

407.143

407.471

1930.8

1943

1940.4

1960.40

1952.43

1969.9


Table A-5 Test Case 5—Root Bridge Down

Run #
Baseline
A to B
Baseline
B to A
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

0

0

0

0

0

0

0

0

2

0

0

0

0

0

0

0

0

3

0

0

0

0

0

0

0

0

Avg

0

0

0

0

0

0

0

0


Table A-6 Test Case 6—Root Bridge Up

Run #
Baseline
A to B
Baseline
B to A
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

173.03

174.24

809.5

795.7

898.2

879.5

923.6

899.9

2

57.83

55.93

863.3

848.7

733.1

715.2

912.5

882.7

3

82.74

80.09

743

726.9

857.5

836.8

853.9

838.3

Avg

104.53

103.42

805.27

790.43

829.6

810.50

896.67

873.63


Table A-7 Test Case 7—Stack Master Down on CZ-C3750

Run #
Baseline
A to B
Baseline
B to A
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

8.3

13.40

16.1

25.6

14.1

28.6

28.7

31.2

2

9.8

15.49

16.3

24.3

14.8

26.6

22.7

28.2

3

8.6

15.40

16.7

23.3

11.1

25.6

25.7

30.2

Avg

8.90

14.76

16.37

24.40

13.33

26.93

25.70

29.87


Table A-8 Test Case 8—Stack Master Re-Established on CZ-C3750

Run #
Baseline
A to B
Baseline
B to A
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

153.10

145.19

982.5

968.6

919.2

898

883.2

865.3

2

155.10

148.19

891.8

859.6

949.2

924.5

984.2

978.2

3

161.39

161.39

851.8

895.6

986.9

974.7

927.3

915.4

Avg

156.53

151.59

908.70

907.93

951.77

932.40

931.57

919.63


Suite 2 Test Cases

Table A-9 Test Case 1—Software Shut Link between Cell-C2955-1 and CZ-C3750

Run #
Baseline
A to B
Baseline
B to A
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

29.60

48.00

43

50.1

51.5

56.5

73.3

95.3

2

32.60

43.49

39.1

46.1

47.2

59.8

68.3

90.7

3

39.70

54.79

45.1

47.2

45

57.4

66.2

88.3

Avg

38.20

48.06

42.4

47.8

47.9

57.9

69.26

91.43


Table A-10 Test Case 2—Software Unshut Link between Cell-C2955-1 and CZ-C3750

Run #
Baseline A to B
Baseline
B to A
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

1275.3

1289.054069

1944.6

1956.2

1923.7

1934.2

1916.8

1942.6

2

1361.1

1377.855617

1954

1965.7

1981.6

1987.5

1999.9

1964.8

3

1469.7

1475.2

1940.1

1951.3

1927.7

1933.8

1909.1

1930.7

Avg

1368.7

1380.703229

1946.233333

1957.733333

1944.333333

1951.833333

1941.933333

1946.033333


Table A-11 Test Case 3—Physically Remove Link between Cell-C2955-1 and CZ-C3750

Run #
Baseline
A to B
Baseline
B to A
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC B to A
200 MAC
A to B
200 MAC
B to A

1

537.4

537.5

474.8

480

570.4

575

492.3

512.3

2

511.6

511.7

526.3

526.1

519.6

529.7

567.2

587.4

3

423.6

423.7

497.6

497.2

492.1

510.2

571.2

576.6

Avg

490.8666667

490.9666667

499.5666667

501.1

527.3666667

538.3

543.5666667

558.7666667


Table A-12 Test Case 4—Physically Re-insert Link between Cell-C2955-1 and CZ-C3750

Run #
Baseline A to B
Baseline
B to A
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

1964.9

1970.8

1983.2

1991.3

1924.5

1938.2

1928.5

1946.4

2

1971.1

1976.9

1965.2

1974.5

1925.3

1939.3

1883.1

1894.5

3

1980.9

1986.5

1950.3

1957.6

1982.5

1962.2

1950.3

1992.1

Avg

1972.3

1978.066667

1966.233333

1974.466667

1944.1

1946.566667

1920.633333

1944.333333


Table A-13 Test Case 5—Root Bridge Down

Run #
Baseline
A to B
Baseline
B to A
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

466.4

466.6

539.1

544.3

580.7

591

859.5

862.3

2

477

477.2

529.1

534.3

797.3

797.8

1034.9

822.6

3

424.4

424.9

556.6

556

790.3

788.4

767.5

787.9

Avg

455.9333333

456.2333333

541.6

544.8666667

722.7666667

725.7333333

887.3

824.2666667


Table A-14 Test Case 6—Root Bridge Up

Run #
Baseline
A to B
Baseline
B to A
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

2280.2

2283.1

2040.4

2192.7

1969.2

1954.6

1931.6

2183

2

1476

1477.8

1973.7

1965.7

1935.9

1918.8

2125

2112.3

3

1979.9

1974.8

2111.5

2099.8

2357.8

2364.4

2013.2

1990.6

Avg

1912.033333

1911.9

2041.866667

2086.066667

2087.633333

2079.266667

2023.266667

2095.3


Table A-15 Test Case 7—Stack Master Down on CZ-C3750

Run #
Baseline
A to B
Baseline
B to A
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

1429.6

1429.9

1740.8

1740.9

1927.3

1927.4

1937.4

1932

2

1650.1

1650.4

1748.6

1759.3

1607.2

1610.6

1646.5

1664.2

3

1139.9

1140.5

1718.8

1719.1

1417.3

1412.6

1346.5

1346.7

Avg

1406.533333

1406.933333

1736.066667

1739.766667

1650.6

1650.2

1643.466667

1647.633333


Table A-16 Test Case 8—Stack Master Re-Established on CZ-C3750

Run #
Baseline
A to B
Baseline
B to A
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

1981.8

1977.2

1948.3

1939.9

1928.4

1923.2

1925.1

1904

2

1978

1983.2

1968.5

1975.6

1935

1935.5

1904.1

1930.5

3

1996.1

1991.6

1952.1

1938.7

1953.2

1926.9

1990.8

1982.7

Avg

1985.3

1984

1956.3

1951.4

1938.866667

1928.533333

1940

1939.066667


Sample Trend Line for Link Failure Between Adjacent Switches

Figure A-6 shows the trend line for link failure between the C2955-7 and C2955-8 switches.

Figure A-6 Link Failure—C2955-7 to C2955-8

Key findings were as follows:

~.5 seconds with one MAC address, and without CIP and producer/consumer traffic

~2 seconds with between 50-200 MAC addresses with producer/consumer traffic

Traffic load (CIP, producer/consumer) increases convergence time

Number of MAC addresses not a large influence on convergence time

Sample Trend Line for Link Failure To Root Bridge

Figure A-7 shows the trend line for link failure to the root bridge.

Figure A-7 Link Failure to Root Bridge

Key findings were as follows:

Convergence time is ~490-590 milliseconds, with a slight upward trend depending on number of MAC addresses.

16-Switch Ring—STP Testing

To verify some scaling parameters, testing was performed with double the amount of 2955s in the cell/area zone (16), and a spot check of certain test cases was performed to compare against the 8-switch ring STP tests. The same traffic flow and test methodology exists from the 8-switch tests except that there are more L2 hops from source to destination. Following are the tests and the corresponding results.

Test Suite 1—Bidirectional Traffic from (Tx1 <-> Tx2)

Table A-17 Test Case 1—Physically Remove Link between Cell-C2955-7 and Cell-C2955-8

Run #
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

1522.6

1522.5

1907.7

1902

3131.4

3131.4

2

1363.8

1363.7

2299.1

2289.2

2872.7

2858.7

3

1843

1834.4

1934.2

1923

2273.8

2273.7

Avg

1576.466667

1573.533333

2047

2038.066667

2759.3

2754.6


Table A-18 Test Case 2—Physically Reinsert Link between Cell-C2955-7 and Cell-C2955-8

Run #
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

954.6

968.9

1993.5

2005.5

1715.4

1743.9

2

1135.2

1149.3

1779.1

1795.2

1987.4

2004.4

3

1993.3

2003.8

1789.5

1805.9

1977.6

2003.2

Avg

1361.033333

1374

1854.033333

1868.866667

1893.466667

1917.166667


Key findings were as follows:

Very similar results as 8 switch tests

Slightly longer convergence times with 200*2 MAC Addresses from ~2.1 seconds to ~2.7 seconds upon link failure

Other tests not performed because of the similarity of the worst case scenario test from above


Note No baseline measurements were gathered (background traffic was always running).


Test Suite 2—Bidirectional Traffic (Tx3 <-> Tx4)

Table A-19 Test Case 1—Physically Remove Link between Cell-C2955-1 and CZ-C3750

Run #
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

585.8

586.2

653.1

645.1

615.7

598.8

2

606

601.1

508.3

500.5

673.8

654.2

3

581.2

576.6

573.6

564.4

648.1

628.4

Avg

591

587.9666667

578.3333333

570

645.8666667

627.1333333


Table A-20 Test Case 2—Physically Reinsert Link between Cell-C2955-1 and CZ-C3750

Run #
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

1809.3

1818.7

1758.9

1777.3

1787.5

1775.5

2

1820.8

1872.5

1783.9

1786.3

1707.8

1680.8

3

1817.2

1827.4

1815.4

1824.2

1795

1820.2

Avg

1815.766667

1839.533333

1786.066667

1795.933333

1763.433333

1758.833333


Key findings were as follows:

Very similar results as 8-switch tests

Slightly longer convergence time on link failure test (all MAC addresses), which was expected

Other tests not performed because of the similarity of the worst-case scenario test from above


Note No baseline measurements were gathered (background traffic was always running).


Redundant Star Topology—STP Testing

Although the majority of the testing was done with the ring topology, the redundant star topology was also tested for comparison purposes. (See Cell/Area Network—Star Topology, page 2-24.) Ixia connections were used between two adjacent 2955 switches. The following test cases were performed.

Table A-21 Test Case 1—Shut Non-Blocking Link on C2955-12

Run #
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

24.5

43.9

35.1

61.9

45.5

88.2

2

23.9

50.3

29.8

65.8

44.9

85.2

3

23

46.5

31.8

59.4

47.2

96.5

Avg

23.8

46.9

32.23333333

62.36666667

45.86666667

89.96666667


Table A-22 Test Case 2—Shut Non-Blocking Link on C2955-12

Run #
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

1999.7

2015

2017.9

2038.4

1986.8

2016.5

2

2000.5

2012.9

1997.7

2015

1992.6

2017.1

3

1998.8

2015.7

1991.3

2011.9

1994.8

2015.6

Avg

1999.666667

2014.533333

2002.3

2021.766667

1991.4

2016.4


Table A-23 Test Case 3—Physically Remove Non-Blocking Link on C2955-12 (Wire Cut Simulation)

Run #
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

555.8

598.6

449.9

478.7

523.5

563.2

2

489.3

510.7

523.5

548.1

479.3

522.8

3

382

404.7

492.7

518.9

509.1

543.6

Avg

475.7

504.6666667

488.7

515.2333333

503.9666667

543.2


Table A-24 Test Case 4—Physically Re-Insert Non-Blocking Link on C2955-12 (Wire Cut Simulation)

Run #
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

2000.7

2016.3

1993.5

2013.5

1990.6

2012.9

2

2004

2017.7

1996

2014.8

1992.8

2014

3

1994.9

2011.1

2001.2

2018.2

1993.9

2016.6

Avg

1999.866667

2015.033333

1996.9

2015.5

1992.433333

2014.5


Table A-25 Test Case 5—Fail Root Bridge (Slot 1 on 3750)

Run #
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

591

591.3

576

501.9

566.3

520

2

554.3

588.9

584.1

545.4

483.9

570.1

3

562.6

603

532.4

569.1

567

595

Avg

569.3

594.4

564.1666667

538.8

539.0666667

561.7


Table A-26 Test Case 5—Re-establish Root Bridge (Slot 1 on 3750)

Run #
50 MAC
A to B
50 MAC
B to A
100 MAC
A to B
100 MAC
B to A
200 MAC
A to B
200 MAC
B to A

1

131.2

197.7

225.7

240.1

323.3

400.1

2

118.5

193.9

210

308.6

298.2

380

3

123.2

196.7

200.2

337.3

311

281.7

Avg

124.3

196.1

211.9666667

295.3333333

310.8333333

353.9333333


Key findings were as follows:

Best convergence times compared to ring-8 or ring-16, averaging 500ms consistently

More consistent numbers

Worse case for traffic flow is only two L2 hops away

Latency/Jitter Testing

To characterize different network topologies under steady state, latency and jitter measurements were captured. Unlike spanning tree convergence testing, no failures were introduced. These tests assume that the network is functioning normally with typical control device traffic running in the background. Simulated source/destination patterns are worse-case scenarios (in the ring topologies) with traffic traversing the entire ring. This is done by having knowledge of the STP-blocked port before the testing begins.

The following test cases were completed for latency/jitter measurements:

8-switch ring—Bidirectional traffic 2955-5 <> 2955-4

16-switch ring—Bidirectional traffic 2955-8 <> 2955-9

Hub/spoke—Bidirectional traffic 2955-12 <> 2955-13

The following results are in microseconds (µ):

Table A-27 Latency/Jitter Test Results

Use Case
Latency
Tx3 >Tx4
Jitter
Tx3 >Tx4
Latency
Tx4 >Tx3
Jitter
Tx4 >Tx3

1

43.068

30.94

42.822

33.8

2

65.902

36.92

65.606

36.94

3

27.022

36.34

25.447

34.7


Key findings were as follows:

Consistent with results from disruptive tests from above

Hub/spoke has the best latency, followed by 8-ring and 16-ring.

Jitter was consistent across all tests.

IGMP Testing

IGMP Snooping Test Methodology

The same eight devices that were used for the STP tests were also used to verify IGMP snooping. The testing exercised various combinations of IGMP snooping with and without a querier on various switches in the network, with active producer-consumer traffic running between PACs and a variable frequency drive. However, as per the Cisco recommended deployment, IGMP snooping works properly only in the presence of a querier. (For more information, see the following URL: http://www.cisco.com/en/US/products/hw/switches/ps708/products_tech_note09186a008059a9df.shtml).

Thus, the only relevant test results are IGMP snooping with querier, and no IGMP snooping with or without querier enabled. A protocol analyzer (http://www.wireshark.org) was used to verify the presence of multicast data traffic.

IGMP Snooping Test Topology

Figure A-8 shows the IGMP snooping test topology.

Figure A-8 IGMP Snooping Test Topology

In this topology, Controller A (Receiver/Consumer) is consuming traffic from Drive A (Producer/Source) across the ring topology. On the same switch that is producing multicast data traffic, a sniffer (Sniffer A) is connected on a different port that has IGMP snooping enabled to verify that this traffic is not seen on this port. The same verification is done on another port with IGMP snooping disabled. Finally, the test was repeated on a different switch in the ring with Sniffer C and Sniffer D, respectively.

IGMP Snooping Test Results

Table A-28 shows the IGMP snooping test results. Note the following:

Yes = Receiving multicast data traffic (destined to 239.x.x.x)

No = Not receiving multicast data traffic (destined to 239.x.x.x)

Table A-28 IGMP Snooping Test Results

Querier/Snooping Enabled
IGMP
Client
Non-IGMP Client Same Switch
Non-IGMP Client Different Switch

OFF/OFF

Yes

Yes

Yes

ON/ON

Yes

No

No