Dell POWEREDGE M1000E - Manual (Page 537)

Fabric OS Administrator’s Guide 537
53-1002745-02
Recommendations for trunk groups
22
Recommendations for trunk groups
To identify the most useful trunk groups, consider the following recommendations along with the
standard guidelines for SAN design:
• Evaluate the traffic patterns within the fabric.
• Place trunking-capable switches adjacent to each other.
This maximizes the number of trunk groups that can form. If you are using a core and edge
topology, place trunking-capable switches at the core of the fabric and any switches that are
not trunking-capable at the edge of the fabric.
• When connecting two switches with two or more ISLs, ensure that all trunking requirements
are met to allow a trunk group to form.
• Determine the optimal number of trunk groups between each set of linked switches,
depending on traffic patterns and port availability.
The goal is to avoid traffic congestion without unnecessarily using ports that could be used to
attach other switches or devices. Consider these points:
- Each physical ISL uses two ports that could otherwise be used to attach node devices or
other switches.
- Trunk groups can be used to resolve ISL oversubscription if the total capability of the trunk
group is not exceeded.
• Consider how the addition of a new path will affect existing traffic patterns:
- A trunk group has the same link cost as the master ISL of the group, regardless of the
number of ISLs in the group. This allows slave ISLs to be added or removed without
causing data to be rerouted, because the link cost remains constant.
- The addition of a path that is shorter than existing paths causes traffic to be rerouted
through that path.
- The addition of a path that is longer than existing paths may not be useful, because the
traffic will choose the shorter paths first.
• Plan for future bandwidth addition to accommodate increased traffic.
For trunk groups over which traffic is likely to increase as business requirements grow,
consider leaving one or two ports in the group available for the future nondisruptive addition of
bandwidth.
• Consider creating redundant trunk groups where additional ports are available or paths are
particularly critical.
This helps to protect against oversubscription of trunk groups, multiple ISL failures in the same
group, and the rare occurrence of an ASIC failure.
• To provide the highest level of reliability, deploy trunk groups in redundant fabrics to help
ensure that ISL failures do not disrupt business operations.