SLB Redundancy Mechanisms
The only redundancy mechanisms used by SLB are link loss and transmit validation heartbeats.
None of the advanced redundancy mechanisms (for example, Active Path and Fast Path) are
supported by SLB. The advanced redundancy mechanisms require the ability to predict and
control which ports in a team will receive a particular frame. Since the switch connected to an
SLB team controls how the team receives frames, the advanced redundancy mechanisms would
not function reliably.
Any SLB teamed port that loses link or fails the transmit validation heartbeat test will be set to
a failed state and will not be used by the team until the issue is resolved.
SLB Network Adapter Failure Recovery
With SLB, the recovery mechanism is somewhat different than those discussed previously. All
members of the team transmit and receive frames with the same source MAC address in the
Ethernet header and there is no concept of a Primary or Non-Primary port as there is in NFT and
TLB. With SLB, there are no receive validation heartbeat frames, and consequently no receive
validation heartbeat failures. In a two-port SLB team, all members are capable of receiving data
frames (based on the switch's load-balancing algorithm), and transmitting data frames (based
on the teaming Driver's load balancing algorithm). In the event of a failover, all transmit traffic
is redistributed among the working teamed ports. After a teamed port failure occurs in a two-port
team, all transmit traffic is sent using a single teamed port since only one port is currently working.
All receive traffic is determined by the switch, which should detect that only one teamed port is
functional. If a failed teamed port is restored, all transmit and receive traffic is once again load
balanced among all teamed ports.
A failure on any teamed port within the same SLB team has the same effect as a failure on any
other teamed port because all teamed ports are considered "equal" (except for link loss or transmit
validation heartbeat failures).
SLB Applications
Switch-assisted Load Balancing is deployed in environments that require fault tolerance and
additional transmit and receive throughput greater than the capacity of the Primary port. SLB
environments also require a switch manually configured to provide Port Trunking.
Recommended Configurations for SLB Environments
HP recommends the following:
•
Transmit validation heartbeats should be enabled (default).
•
MAC addresses should not be manually set to a locally administered address (LAA) via the
Microsoft Adapter Properties User Interface. A user should not implement LAAs on
individual network ports that are members of a team; otherwise teaming may not function
correctly. Setting an LAA for the team is permitted via the NCU.
•
Spanning Tree's blocking, listening, and learning stages should be disabled or bypassed on
all switch ports to which an HP Integrity Network adapter team port is attached. These
stages are not needed when a non-switch networking device (for example, server) is attached
to the switch port. HP ProCurve switches have a feature called STP Fast Mode that is used
to disable these Spanning Tree stages on a port-by-port basis. Cisco switches have an
equivalent feature called PortFast.
•
SLB teams that communicate with TCP/IP network devices via a router should use the
Automatic, TCP port or the IP address-based load-balancing algorithm (configured via the
NCU).
•
Implementers thoroughly understand the configuration guidelines set by the switch vendor
because SLB is dependent on the switch being configured in a compatible mode. HP's SLB
72
The Mechanics of Teaming for the Advanced User