HP Integrity rx2800 HP Integrity Network Adapter Teaming Whitepaper - Page 39

Directed ARP, Community Address ARP, Active Path Directed ARP, Directed ARP's Echo Node Probe

Page 39 highlights

the Echo Node will always resolve the team's IP address to a single port in the team, which would prevent Echo Node probe responses from being sent to every port in the team. Active Path uses a special frame as the Echo Node probe. This frame is identical to a normal ARP request frame except for one field - the Sender's IP address field. In an Echo Node probe, the sender's IP address field is intentionally left blank. This frame format accomplishes three goals: 1) the probe provides the MAC address that the Echo Node should respond to (in the sender's Hardware Address field), 2) the probe does not provide a sender's IP address so that the ARP cache on the Echo Node is not affected by the Echo Node probe, and 3) by using a modified ARP request frame, the Echo Node does not require any type of special configuration other than the assignment of an IP address. Note: Not all TCP/IP stacks will respond to the Directed ARP Echo Node probe. When Active Path is enabled on a team, an Echo Node probe is transmitted from every member port in the team every X seconds (where X is equal to the setting of the Echo Node probe interval). The Active Path mechanism expects to receive a response on every member port every Y number of seconds (where Y is equal to the setting of the Echo Node probe timeout). Once a particular member port fails to receive an Echo Node probe response, the port is considered to be in a failed/disabled state. The port continues to transmit Echo Node probes every X number of seconds but will not be re-enabled for use until at least one Echo Node probe response is received on the port. The default values for Echo Node probe interval (X) and Echo Node probe timeout (Y) are both 3 seconds. If the Echo Node is unreachable or unavailable for all ports in a team, the team considers all member ports equal and will not disable any ports based on the Active Path mechanism. Another mechanism would have to cause a member port failure. As long as at least one member port can reach the Echo Node, failure to communicate with the Echo Node for any other port will cause the port to be disabled for use by the team. There are two Active Path mechanisms: Directed ARP and Community Address ARP. • Active Path Directed ARP - Active Path's Directed ARP can be described as a "one-to-one" validation mechanism between each port in a team and the designated Echo Node. With Directed ARP, each and every teamed port transmits an Echo Node probe request to the designated Echo Node and expects an individual Echo Node reply in response. The Echo Node probe requests are sent at the interval configured in the teaming GUI in the Echo Node Probe Interval field. Each teamed port that transmits an Echo Node probe request will wait for the number of Echo Node probe timeout seconds (configured in the teaming GUI). If the Echo Node probe request times out without receiving a response from the Echo Node, the teaming driver will assume the teamed port does not have connectivity with the Echo Node and will place the port in Active Path failure. This prevents the port from being used as the Primary port in the team and will prevent the team from using it as a Secondary (non-Primary) in load balancing teams (for example, TLB). If the Primary port in a team loses connectivity with the Echo Node, another port in the team with confirmed connectivity with the Echo Node will assume the role of Primary (resulting in a failover). Directed ARP's Echo Node Probe: Special considerations were required when choosing a frame type to use for individually validating every teamed port's connectivity with an external device (Echo Node). First, the frame type used by Echo Node had to allow for a teamed port to predictably receive a response from the Echo Node to validate its individual connectivity. Since the team's IP address resolves to a specific MAC address that's associated with the team's Primary port, normal ICMP (in other words, ping) frames could not be used. If ICMP Requests were transmitted by each teamed port, the responses would all be transmitted to the Primary port (programmed with the team's MAC address). If the Primary port was isolated from the Echo Node, no responses would be received for any teamed ports. As a result, Active Path would be unable to determine which teamed ports did or didn't have connectivity with the Echo Node and a failure recovery would not be possible. Second, the frame type had to be supported by as many IP protocol stacks as possible, so no special configuration or Types of HP Integrity Network Adapter Teams 39

  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 10
  • 11
  • 12
  • 13
  • 14
  • 15
  • 16
  • 17
  • 18
  • 19
  • 20
  • 21
  • 22
  • 23
  • 24
  • 25
  • 26
  • 27
  • 28
  • 29
  • 30
  • 31
  • 32
  • 33
  • 34
  • 35
  • 36
  • 37
  • 38
  • 39
  • 40
  • 41
  • 42
  • 43
  • 44
  • 45
  • 46
  • 47
  • 48
  • 49
  • 50
  • 51
  • 52
  • 53
  • 54
  • 55
  • 56
  • 57
  • 58
  • 59
  • 60
  • 61
  • 62
  • 63
  • 64
  • 65
  • 66
  • 67
  • 68
  • 69
  • 70
  • 71
  • 72
  • 73
  • 74
  • 75
  • 76
  • 77
  • 78
  • 79
  • 80
  • 81
  • 82
  • 83
  • 84
  • 85
  • 86
  • 87
  • 88
  • 89
  • 90
  • 91
  • 92
  • 93
  • 94
  • 95
  • 96
  • 97
  • 98
  • 99
  • 100
  • 101
  • 102
  • 103

the Echo Node will always resolve the team’s IP address to a single port in the team, which
would prevent Echo Node probe responses from being sent to every port in the team.
Active Path uses a special frame as the Echo Node probe. This frame is identical to a normal ARP
request frame except for one field – the Sender’s IP address field. In an Echo Node probe, the
sender’s IP address field is intentionally left blank. This frame format accomplishes three goals:
1) the probe provides the MAC address that the Echo Node should respond to (in the sender’s
Hardware Address field), 2) the probe does not provide a sender’s IP address so that the ARP
cache on the Echo Node is not affected by the Echo Node probe, and 3) by using a modified ARP
request frame, the Echo Node does not require any type of special configuration other than the
assignment of an IP address. Note: Not all TCP/IP stacks will respond to the Directed ARP Echo
Node probe.
When Active Path is enabled on a team, an Echo Node probe is transmitted from every member
port in the team every X seconds (where X is equal to the setting of the Echo Node probe interval).
The Active Path mechanism expects to receive a response on every member port every Y number
of seconds (where Y is equal to the setting of the Echo Node probe timeout). Once a particular
member port fails to receive an Echo Node probe response, the port is considered to be in a
failed/disabled state. The port continues to transmit Echo Node probes every X number of seconds
but will not be re-enabled for use until at least one Echo Node probe response is received on the
port. The default values for Echo Node probe interval (X) and Echo Node probe timeout (Y) are
both 3 seconds.
If the Echo Node is unreachable or unavailable for all ports in a team, the team considers all
member ports equal and will not disable any ports based on the Active Path mechanism. Another
mechanism would have to cause a member port failure. As long as at least one member port can
reach the Echo Node, failure to communicate with the Echo Node for any other port will cause
the port to be disabled for use by the team.
There are two Active Path mechanisms:
Directed ARP
and
Community Address ARP
.
Active Path Directed ARP
– Active Path’s Directed ARP can be described as a “one-to-one”
validation mechanism between each port in a team and the designated Echo Node. With
Directed ARP, each and every teamed port transmits an Echo Node probe request to the
designated Echo Node and expects an individual Echo Node reply in response. The Echo
Node probe requests are sent at the interval configured in the teaming GUI in the Echo Node
Probe Interval field. Each teamed port that transmits an Echo Node probe request will wait
for the number of Echo Node probe timeout seconds (configured in the teaming GUI). If the
Echo Node probe request times out without receiving a response from the Echo Node, the
teaming driver will assume the teamed port does not have connectivity with the Echo Node
and will place the port in Active Path failure. This prevents the port from being used as the
Primary port in the team and will prevent the team from using it as a Secondary
(non-Primary) in load balancing teams (for example, TLB). If the Primary port in a team
loses connectivity with the Echo Node, another port in the team with confirmed connectivity
with the Echo Node will assume the role of Primary (resulting in a failover).
Directed ARP’s Echo Node Probe:
Special considerations were required when choosing a frame type to use for individually
validating every teamed port’s connectivity with an external device (Echo Node). First, the
frame type used by Echo Node had to allow for a teamed port to predictably receive a
response from the Echo Node to validate its individual connectivity. Since the team’s IP
address resolves to a specific MAC address that’s associated with the team’s Primary port,
normal ICMP (in other words, ping) frames could not be used. If ICMP Requests were
transmitted by each teamed port, the responses would all be transmitted to the Primary port
(programmed with the team’s MAC address). If the Primary port was isolated from the Echo
Node, no responses would be received for any teamed ports. As a result, Active Path would
be unable to determine which teamed ports did or didn’t have connectivity with the Echo
Node and a failure recovery would not be possible. Second, the frame type had to be
supported by as many IP protocol stacks as possible, so no special configuration or
Types of HP Integrity Network Adapter Teams
39