Seen our Load Balancing Algorithms Page but still want more detail? This post is for you. Here are additional details regarding each algorithmn taken from one of our previous support documents:
Weighted Balance
This setting specifies the ratio of WAN connection usage to be applied on the specified IP protocol and port. This setting is applicable only when Algorithm is set to Weighted Balance.
The amount of matching traffic that is distributed to a WAN connection is proportional to the weight of the WAN connection relative to the total weight. Use the sliders to change each WAN’s weight.
For example, with the following weight settings:
For example, with the following weight settings on a Peplink Balance 310:
WAN1: 10
WAN2: 10
WAN3: 5
Total weight is 25 = (10 + 10 + 5)
Matching traffic distributed to WAN1 is 40% = (10 / 25) x 100%.
Matching traffic distributed to WAN2 is 40% = (10 / 25) x 100%.
Matching traffic distributed to WAN3 is 20% = (5 / 25) x 100%.
Algorithm: Persistence
The configuration of persistent services is the solution to the few situations where link load distribution for Internet services is undesirable. For example, for security reasons, many e-banking and other secure websites terminate the session when the client computer’s Internet IP address changes mid-session.
In general, different Internet IP addresses represent different computers. The security concern is that an IP address change during a session may be the result of an unauthorized intrusion attempt. Therefore, to prevent damages from the potential intrusion, the session is terminated upon the detection of an IP address change.
Pepwave routers can be configured to distribute data traffic across multiple WAN connections. Also, the Internet IP depends on the WAN connections over which communication actually takes place. As a result, a LAN client computer behind the Pepwave router may communicate using multiple Internet IP addresses. For example, a LAN client computer behind a Pepwave router with three WAN connections may communicate on the Internet using three different IP addresses.
With the persistence feature, rules can be configured to enable client computers to persistently utilize the same WAN connections for e-banking and other secure websites. As a result, a client computer will communicate using one IP address, eliminating the issues mentioned above.
There are two persistent modes: By Source and By Destination.
By Source:
The same WAN connection will be used for traffic matching the rule and originating from the same machine, regardless of its destination. This option will provide the highest level of application compatibility.
By Destination:
The same WAN connection will be used for traffic matching the rule, originating from the same machine, and going to the same destination. This option can better distribute loads to WAN connections when there are only a few client machines.
The default mode is By Source. When there are multiple client requests, they can be distributed (persistently) to WAN connections with a weight. If you choose Auto in Load Distribution, the weights will be automatically adjusted according to each WAN’s Downstream Bandwidth which is specified in the WAN settings page). If you choose Custom, you can customize the weight of each WAN manually by using the sliders.
Algorithm: Enforced
This setting specifies the WAN connection usage to be applied on the specified IP protocol and port. This setting is applicable only when Algorithm is set to Enforced.
Matching traffic will be routed through the specified WAN connection, regardless of the health check status of the WAN connection. Starting from Firmware 5.2, outbound traffic can be enforced to go through a specified SpeedFusion connection.
Algorithm: Priority
This setting specifies the priority of the WAN connections used to route the specified network service. The highest priority WAN connection available will always be used for routing the specified type of traffic. A lower priority WAN connection will be used only when all higher priority connections have become unavailable.
Outbound traffic can be prioritized to go through SpeedFusion connection(s). By default, VPN connections are not included in the priority list.
Algorithm: Overflow
The traffic matching this rule will be routed through the healthy WAN connection that has the highest priority and is not in full load. When this connection gets saturated, new sessions will be routed to the next healthy WAN connection that is not in full load.
Drag and drop to specify the order of WAN connections to be used for routing traffic. Only the highest priority healthy connection that is not in full load will be used.
Algorithm: Least Used
The traffic matching this rule will be routed through the healthy WAN connection that is selected in Connection and has the most available download bandwidth. The available download bandwidth of a WAN connection is calculated from the total download bandwidth specified on the WAN settings page and the current download usage. The available bandwidth and WAN selection is determined every time an IP session is made.
Algorithm: Lowest Latency
The traffic matching this rule will be routed through the healthy WAN connection that is selected in Connection and has the lowest latency. Latency checking packets are issued periodically to a nearby router of each WAN connection to determine its latency value. The latency of a WAN is the packet round trip time of the WAN connection. Additional network usage may be incurred as a result.
The roundtrip time of a 6M down/640k uplink can be higher than that of a 2M down/2M up link because the overall round trip time is lengthened by its slower upload bandwidth, despite its higher downlink speed. Therefore, this algorithm is good for two scenarios:
- All WAN connections are symmetric; or
- A latency sensitive application must be routed through the lowest latency WAN, regardless of the WAN’s available bandwidth.