ArubaOS 8.6.0.0 Help Center
You are here: Home > Cluster > Load Balancing

Cluster Load Balancing

Cluster load-balancing is achieved through the features, Client load-balancing and AP load-balancing. Both these features are explained in this section.

Client Load Balancing

The client load balancing feature ensures that clients are evenly distributed across the cluster members, thereby using the system resources efficiently.

If the system detects a distorted distribution of load, it balances the load on the managed devices by changing the UAC of these clients. The load across all the managed devices is balanced in the cluster regardless of the type of platform.

The cluster manager calculates the ratio of the existing number of clients on a managed device and its maximum capacity. Based on this ratio and additional threshold triggers, client load balancing is triggered.

When any new managed device, including the managed device that comes up after a failover, is added to an existing cluster, it is considered for load balancing and accordingly, APs and clients are moved to balance the load in the cluster.

Load balancing is enabled by default when a cluster is configured.

Threshold triggers

  • Active client rebalance threshold: The actual active load on a cluster member. The threshold is set at 20%, that is, 20% of the capacity of a platform.
  • Standby client rebalance threshold: The standby load on a cluster member. The threshold is set at 40%.
  • Unbalanced threshold: The difference between the loads on maximum loaded cluster node and the minimum loaded cluster node. The threshold is set at 5%, that is, there must be at least a 5% disparity in load between the managed devices.
  • AP Total Load Balance threshold: The total load balance threshold is set to 40%. This is the default value and cannot be configured.

For load balancing to be triggered for active clients, the active client rebalance threshold and the unbalanced threshold percentages must be met. Similarly, for the standby client, the standby client rebalance threshold and the unbalanced threshold percentages must be met.

When the redundancy mode is enabled, the capacity of the cluster is reduced to half.

AP Load Balancing

The AP load balancing feature ensures that the cluster leader manages the load balancing based on the platform capacity. The AP is dynamically assigned an AAC when it connects to a cluster. Here, instead of client load, AP load is considered.

Both active and standby APs are considered for load balancing.

Following is the AP load balancing criteria if a managed device is newly added:

  • When an AP threshold is already met in the cluster nodes, if a new managed device is added, the Active AP table of the new managed device is filled first based on AP count set.
  • When the threshold is not met, APs are moved to standby AP table of the newly added managed device.
  • The count of these APs will increment based on the AP count set only after the stabilization of the cluster, however, the APs that were moved during this phase cannot be always based on AP Count.

Starting from ArubaOS 8.3.0.0, the Active AP load balancing feature is enabled by default. In previous releases, this feature is disabled by default.

The active AP load balancing is performed using VRRPVirtual Router Redundancy Protocol. VRRP is an election protocol that dynamically assigns responsibility for a virtual router to one of the VRRP routers on a LAN. for L2 connection and the switch IP of one cluster member for L3 connection. The AP Total Load Balance threshold is set to 40% and active AP load balancing threshold value is set to 20%. This is the default value.

Prior to ArubaOS 8.3.0.0, the cluster leader considered the AP load on each cluster member and assigned the cluster member with least total AP load as the AAC.

For initial load balancing, cluster leader evaluates if the managed device with the least active AP load percentage can accommodate additional APs. If yes, return this managed device as the candidate AAC.

For periodic load balancing, the cluster leader performs load balancing based on the following conditions:

  1. Finds the managed devices with maximum and minimum active load percentage.
  2. Finds the managed devices with maximum and minimum total load percentage
  3. Checks if the max active load percentage is more than the active load percentage threshold. Also, checks if the difference between the maximum and minimum loaded managed devices is more than the unbalance threshold.
  4. Moves the active APs from maximum loaded managed devices to minimum loaded managed device. However, if it is unable to move the Active APs, it will re-balance the standby load by moving the standby APs from the maximum load percentage managed device to minimum load percentage managed device.

The periodic load re-balancing occurs every 1 minute, which is the default value and APs are considered for load re-balancing based on AP count, which is 50 by default.

After APs are listed on a cluster member, the cluster manager periodically recalculates the load of the cluster members to balance the load. For example, when a new managed device joins the cluster.

The triggers for client load balancing and AP load balancing are same.

Listed below are the advantages of AP load balancing

Configuring Cluster Load Balancing

The following procedure describes how to configure load balancing for a cluster:

  1. In the Managed Network node hierarchy, navigate to the Configuration > Services > Clusters tab.
  2. In the Clusters table, select a cluster to configure AP load balancing.
  3. In the Cluster Profile > <cluster name> window, expand Advanced.
  4. Configure the active AP load balancing settings described in Table 1.
  5. Click Submit.
  6. Click Pending Changes.
  7. In the Pending Changes window, select the check box and click Deploy changes.

Table 1: Active AP Load Balancing Parameters

Parameter

Description

Redundancy

Enable redundancy. When the redundancy mode is enabled, the capacity of the cluster is reduced to half and only 8000 clients are considered to reach the threshold.

Active client rebalance threshold

Indicates the minimum total load percentage required to perform AP load balancing in the cluster. At least one of the managed devices must have this value as the total load percentage to trigger AP load balancing.

Standby client rebalance threshold

Indicates the minimum total load percentage required to perform standby AP load balancing in the cluster.

Unbalance threshold

If a managed device reaches the total load threshold, AP load balancing is triggered. This happens when the difference between the loads on the maximum loaded managed device and the minimum loaded managed device in the cluster exceeds the unbalance threshold value.

Heartbeat threshold

Minimum heartbeat threshold is set in milliseconds. The default setting is based on the latency determined between each pair of managed devices and the cluster. It also depends on the connection type between managed device and distribution switch (single ethernet cable, or port channel, and so on).

The following procedure describes how to configure the AP and client load balancing values:

  1. In the Managed Network node hierarchy, navigate to the Configuration > System tab.
  2. Click Profiles. Expand Cluster and click Classic Controller Cluster.
  3. Click + to create a cluster profile.
  4. Add the AP load balancing values and Client Load Balancing values.
  5. Click Submit.
  6. Click Pending Changes.
  7. In the Pending Changes window, select the check box and click Deploy changes.

The following CLICommand-Line Interface. A console interface with a command line shell that allows users to execute text input as commands and convert these commands to appropriate functions. commands configure load balancing for a cluster:

Following is an example of Active AP load balancing:

(7210-24) #show lc-cluster load distribution ap

Cluster Load Distribution for APs

---------------------------------

Type IPv4 Address Active APs Standby APs

---- --------------- -------------- ---------------

peer 192.168.28.23 25 20

self 192.168.28.24 20 25

 

Total: Active APs 45 Standby APs 45

 

(host) #show lc-cluster group-membership

Cluster Enabled, Profile Name = "ap-lb"

Redundancy Mode On

Active Client Rebalance Threshold = 20%

Standby Client Rebalance Threshold = 40%

Unbalance Threshold = 5%

AP Load Balancing: Enabled

Active AP Rebalance Threshold = 20%

Active AP Unbalance Threshold = 5%

Active AP Rebalance AP Count = 50

Active AP Rebalance Timer = 1 minutes

Cluster Info Table

------------------

Type IPv4 Address Priority Connection-Type STATUS

---- --------------- -------- --------------- ------

peer 192.168.28.23 128 L2-Connected CONNECTED (Leader, last HBT_RSP 10ms ago, RTD = 0.000 ms)

self 192.168.28.24 128 N/A CONNECTED (Member)

 

To check if Active AP load balancing is enabled or disabled:

Example for IPv4:

(host) #show lc-cluster group-membership

Cluster Enabled, Profile Name = "testLB"

Redundancy Mode On

Active Client Rebalance Threshold = 20%

Standby Client Rebalance Threshold = 40%

Unbalance Threshold = 5%

AP Load Balancing: Enabled

Active AP Rebalance Threshold = 20%

Active AP Unbalance Threshold = 5%

Active AP Rebalance AP Count = 50

Active AP Rebalance Timer = 5 minutes

Cluster Info Table

------------------

Type IPv4 Address Priority Connection-Type STATUS

---- --------------- -------- --------------- ------

self 192.168.10.38 128 N/A CONNECTED (Leader)

peer 192.168.10.34 128 L2-Connected CONNECTED (Member, last HBT_RSP 38ms ago, RTD= 0.000 ms)

 

Example for IPv6:

(host) #show lc-cluster group-membership

Cluster Enabled, Profile Name = "72xx"

Redundancy Mode On

Active Client Rebalance Threshold = 20%

Standby Client Rebalance Threshold = 40%

Unbalance Threshold = 5%

AP Load Balancing: Enabled

Active AP Rebalance Threshold = 20%

Active AP Unbalance Threshold = 40%

Active AP Rebalance AP Count = 50

Active AP Rebalance Timer = 1 minutes

Cluster Info Table

------------------

Type IPv6 Address Priority Connection-Type STATUS

---- ------------------- --------- --------------- -------------------------

peer 2000:192:168:28::24 128 L2-Connected CONNECTED (Member,last HBT_RSP 68ms ago, RTD = 0.000 ms)

peer 2000:192:168:28::26 128 L2-Connected CONNECTED (Member,last HBT_RSP 66ms ago, RTD = 0.000 ms)

peer 2000:192:168:28::22 128 L2-Connected CONNECTED (Member,last HBT_RSP 69ms ago, RTD = 0.503 ms)

self 2000:192:168:28::23 128 N/A CONNECTED (Leader)

To display the number of times a managed device published APs on the AP channel for A-AAC assignment:

(host) #show ap debug gsm-counters

STM GSM Counters

----------------

Name Value

---- -----

AP Publish Events 93

AP Delete Events 29

AP Publish Events(Load Balance)2

AP Delete Events(Load Balance) 2

Radio Publish, Activate, Activate Errors 28 11 0

Radio Delete Events 3

Radio Delete Errors 15

BSS Publish Events 41

Responses to BSS Rcvd 41

BSS Delete Events 18

BSS Delete Errors 30

BSS Delete Key Not Found (included above) 30

STA Publish Events 0

STA Delete Events 0

STA Activate on S-UAC, Errors 0 0

STA Activate for Delete, Errors 0 0

WIRED_AP Publish Events 0

To display the number of times the AP tried to connect to the cluster leader and needed the cluster leader to assign an A-AAC for this AP:

(host) #show lc-cluster gsm counters | exclude 0

Cluster GSM Channel Counters

----------------------------

AP Channel: Adds >> 1

AP Channel: Deletes >> 1

BSS Channel: Adds >> 2

BSS Channel: Section Update >> 2

AP Channel: Adds and Need AAC Assignment>> 1

AP Channel: Deletes from SAPM, AP redirected>> 1

/*]]>*/