Limitations and Compatible Characteristics of Ethernet Link Bundles
This list describes the properties and limitations of ethernet link bundles:
-
The router supports mixed speed bundles. Mixed speed bundles allow member links of different bandwidth to be configured as active members in a single bundle. The ratio of the bandwidth for bundle members must not exceed 10. Also, the total weight of the bundle must not exceed 64. For example, 100Gbps link and 10Gbps links can be active members in a bundle and load-balancing on member links is based on bandwidth weightage.
-
The weight of each bundle member is the ratio of its bandwidth to the lowest bandwidth member. Total weight of the bundle is the sum of weights or relative bandwidth of each bundle member. Since the weight for a bundle member is greater than or equal to 1 and less than or equal to 10, the total member of links in a bundle is less than 64 in mixed bundle case.
-
Any type of Ethernet interfaces can be bundled, with or without the use of LACP (Link Aggregation Control Protocol).
-
A single router can support a maximum of 256 bundle interfaces. Link bundles of only physical interfaces are supported.
-
When enabling HQoS profile, the maximum available trunks by default (bundle main + sub-interfaces) are 256. If you need more trunks, configure the hw-module profile bundle-scale <256/512/1024> command. With HQoS enabled on bundle interfaces, the maximum priority level supported is 4.
-
The following limitations apply to the number of supported bundle members with HQoS profile on Layer2 and Layer3 intefaces:
-
Maximum of 1024 trunks (128 physical interfaces + 896 sub-interfaces) and 16 bundle members.
-
Maximum of 256 trunks (128 physical interfaces + 128 sub-interfaces) and 64 bundle members.
-
Maximum of 512 trunks (128 physical interfaces + 384 sub-interfaces) and 32 bundle members.
-
-
The following limitations apply to bundle sub-interfaces and the number of members per bundle :
-
Maximum of 1024 bundle sub-interfaces, each containing up to 16 member-links.
-
Maximum of 256 bundle sub-interfaces, each containing up to 64 member-links
-
Maximum of 512 bundle sub-interfaces, each containing up to 32 member-links
-
-
Physical layer and link layer configuration are performed on individual member links of a bundle.
-
Configuration of network layer protocols and higher layer applications is performed on the bundle itself.
-
IPv4 and IPv6 addressing is supported on ethernet link bundles.
-
A bundle can be administratively enabled or disabled.
-
Each individual link within a bundle can be administratively enabled or disabled.
-
Ethernet link bundles are created in the same way as Ethernet channels, where the user enters the same configuration on both end systems.
-
QoS is supported and is applied proportionally on each bundle member.
-
In case static MAC address is configured on a bundle-ether interface, the following limitations are applied: -
Locally generated packets, such as ICMP, BGP, and so on, going out from the interface have the source MAC address as the statically configured MAC address.
-
Transit (forwarded) packets going out of the interface do not have the configured static MAC as source MAC address. In such a scenario, the upper 36-bits come from the system MAC address (or the original/dynamic MAC address) and the lower 12-bits come from the MAC address configured on the bundle. To check the dynamic pool of MAC addresses included, use the
show ethernet mac-allocation detail
command.For example, if the dynamic MAC address was 008A.9624.48D8 and the configured static MAC address is 0011.2222.ABCD. Then, the source MAC for transit (forwarded) traffic will be 008A.9624.4BCD.
Note
This limitation can cause traffic blackholing for the transit traffic, in case there is L2 ACL applied for security purpose. In such case, it is necessary to add permit statement for both MAC addresses in the L2 ACL.
-
-
Load balancing (the distribution of data between member links) is done by flow instead of by packet. Data is distributed to a link in proportion to the bandwidth of the link in relation to its bundle.
-
All links within a single bundle must terminate on the same two systems.
-
Bundled interfaces are point-to-point.
-
A link must be in the up state before it can be in distributing state in a bundle.
-
Only physical links can be bundle members.
-
Multicast traffic is load balanced over the members of a bundle. For a given flow, the internal processes selects the member link, and the traffic for the flow is sent over that member.