Physical Link Aggregation—Improving Link Utilization

With high-order QAM and header compression, no further optimization seemed possible in microwave link utilization. However, to meet the ever-increasing needs for larger bandwidth, a simple solution is to establish more links.

The method of combining multiple microwave links in parallel in order to provide a larger throughput is called link aggregation. The primary challenge in link aggregation is how to balance loads among member links.

The traditional 802.1ad-based air interface LAG uses the Hash algorithm to distribute service packets to member links. This scheme is applicable to data-intensive networks with numerous service flows. Generally, the Hash algorithm depends on MAC addresses, IP addresses, MPLS labels, or their combinations to work.

In most cases, the Hash algorithm can effectively balance loads among member links, but it does not work for every link. In some cases, packet loss may occur even though the service traffic does not exceed the maximum LAG bandwidth.

The following examples use vehicle scheduling to illustrate how the Hash algorithm balances loads among member links. The LAG (Hash) algorithm schedules service packets (or, vehicles in this example) based on service packet type (cargo type), it does not consider link utilization (road utilization). Obviously, road resources cannot be fully utilized in this manner.

To fully utilize road resources, vehicle scheduling should be based on the traffic capacity (bandwidth) of each road instead of what type of goods a truck carries. This idea later evolves to the physical link aggregation (PLA) algorithm.

PLA distributes loads according to the bandwidth of each member link. As it does not depend on compositions of L2 and L3 service flows, PLA is also called L1 LAG.

PLA not only can aggregate microwave links carrying frames of different Ethernet frame type and length. It can also combine microwave links of different bandwidth. Moreover, PLA is also able to ensure near-equal Ethernet bandwidth utilization rates on member links when the Ethernet bandwidth of each member link changes as a result of adaptive modulation.

There are three PLA modes:

• PLA: The IF board allocates traffic between two microwave links.
• EPLA: The packet switching unit allocates traffic among a maximum of four microwave links.
• EPLA+: The packet switching unit allocates traffic to the dual-channel IF board which then forwards traffic to IF ports.

All Huawei RTN 300 and RTN 900 products support physical link aggregation. Among them, RTN 980 and RTN 980L support aggregation of up to 12 links. The maximum capacity of the aggregation group reaches an impressive 7 Git/s, providing backhaul bandwidth comparable to optical fiber.

 

===Link archives (updated 2015-05-26)===

>>>Click here for the latest links

 

Amazing Microwaves	Demystifying MSTP	WDM Optical Transport
Huawei RTN: Birth and Evolution	Prelude●Overview and History	Speaking of 100G
How to Maximize BW of Air Interface (1)	MTU, MFL and Jumbo Frames
How to Maximize BW of Air Interface (2)	Key Technologies of EoS: GFP, VCAT and LCAS
How to Maximize BW of Air Interface (3)	PCM Ports, Boards, Services, and Application Scenarios

The best feature Huawei came out with

This feature applies for point to point link, it will be more efficient if it can be for multiple hope link i.e you have two routes between A and B,

Route 1: direct link A-B

Route 2: two links via point C: A-C-B

Its required to balance the traffic between A and B!

 

This could be labeled as EPLA++

 

Regards

Kamal

Dear kamalf,
Thank you so much for your feedback.
I told the post owner and he will reply you later.

Dear kamalf,
Thank you for providing such exciting suggestion. It could be a network level traffic balancing technology, more general than link aggregation.

I except it could be realized, though there must be a lot of work to do, such as link status, different delay of packet transmission, etc. 

 

Best regards!