Configuring WAN Policies in Micro Edge

Overview

WAN policies define how to route outgoing traffic to optimize application performance across multiple paths. Each policy specifies which Internet path to use based on a set of performance metrics or specific criteria. A WAN Policy becomes active when it is set as an action in a WAN Rule.

Example WAN Policies:

Use the WAN with the lowest latency.
Balance across all WANs and favor the WAN with the most available bandwidth.
Balance evenly among the WANs that have less than 100ms latency.
Use a specific WAN such as a VPN Tunnel.

When assigning WAN Rules to the above WAN Policy examples, you can configure Micro Edge to optimize bandwidth as follows:

For VoIP traffic use the WAN with the lowest latency.
For Skype application balance across all WANs and favor the WAN with the most available bandwidth.
For other web traffic balance evenly among the WANs that have less than 100 ms latency.
For HTTP and HTTPS traffic always use a specific VPN Tunnel.

To configure WAN Policies, go to Settings > Routing > WAN Policies.

WAN Policy Parameters

General Parameters

A WAN Policy consists of the following parameters:

Description

A description of the policy. This can be anything you like that identifies the policy & its purpose.

Enabled

Whether to enable or disable the policy. When disabled, any rule that uses the policy is skipped.

Type

Specific WAN - Use a specific WAN
Best WAN - Use the Best WAN (as defined by other settings)
Balance - Balance across multiple WANs (as defined by other settings)

WAN Type Parameters

The Best WAN and Balance types require additional settings:

Best WAN

Metric - the metric used to identify which WAN is the "best" at the time a rule is evaluated.

Lowest Latency - the WAN with the lowest current latency.
Highest Available Bandwidth - the WAN with the highest current available bandwidth.
Lowest Jitter - the WAN with the lowest jitter (most consistent latency).
Lowest Packet Loss - the WAN with the lowest packet loss.

WANs - This defines the set from which the best WAN is chosen.

All WANs - the best WAN from all available and online WANs.
Specific WANs - allows for manual selection of a set of WANs.

Balance

Algorithm - the algorithm used to balance across multiple WANs.

Weighted - balance across the WANs using the configured static weights.
Latency - balance across the WANs favoring the lowest latency WAN.
Available Bandwidth - balance across the WANs favoring the WAN with the most available bandwidth.
Bandwidth - balance across the WANs favoring the WAN with the most bandwidth.

WANs - This defines the set of WANs from which to balance.

All WANs - balance among all WANs.
Specific WANs - balance among only the specified WANs.

WAN Type Criteria

Both "Balance" and "Best WAN" type WAN policies support criteria. By adding criteria you can set "Service Level Agreements" so that any WAN in the policy is dynamically removed (made unavailable) at the time of WAN policy processing.

For example, a policy to balance among all WANs can include criteria to balance among all WANs with less than 100ms latency.

There are several types of criteria:

Attribute

Is VPN: Apply the policy only to VPN interfaces.
Name: Can be used with custom interface names to only use certain WANs based on name. For example, only use WANs that contain the word "Corporate" or "Cable".

Custom metric

Latency - limit the available WAN by latency - units in milliseconds (Example: <100ms). This is useful for traffic that only behaves well on certain low-latency links. If the network has a large amount of voice traffic, we likely want to balance among all WANs that have a certain level of low latency.
Available Bandwidth - limit the available WANs by a certain amount of available bandwidth (Example: >10%). This is useful for removing interfaces from the balancing pool that are saturated. For example, balance among all WANs that have more than 1 Mbit of available bandwidth. If a backup task is always put on a certain WAN using a "Specific WAN" policy, that WAN may be saturated and perform poorly. In this case it can be removed with this criteria.
Jitter - limit the available WAN by jitter - units in milliseconds (Example: <10ms). This is useful to remove WANs from consideration with high jitter which can mean inconsistent behavior. This can be useful for voice traffic which should typically be put on well performing links.
Packet Loss - limit the available WANs by packet loss - units in percent (Example: <1%). This is useful to remove WANs from consideration when some amount of packet loss can mean inconsistent behavior. This can be useful for voice traffic which should typically be put on well performing links.

Connectivity Test

Can be used to "rule out" interfaces which do not have active internet connectivity.

Type - The kind of test used: ping, ARP, DNS, or HTTP request.
Interval - How often is the test conducted? Value is in seconds.
Timeout - How long to wait for a response before the test is considered a failure? Value is in seconds.
Failure Threshold - How many individual failures must be accumulated before the WAN is considered to be down. Higher values require more failures and are less sensitive.
Target - The host or IP address to test against. This should be an external resource that will respond to the selected type of test. For example, Google DNS (8.8.8.8) has functionally 100% uptime and will always respond to ping tests.

WAN Failover

It is possible that based on the constraints of a WAN policy, there is no suitable route. For example, a WAN policy may be set to use a Best WAN, however at the time of processing, all WAN links enabled in the policy are down. In this case, the policy is considered defunct.

When a policy is defunct, all WAN Rules that refer to the policy are skipped. In other words, WAN Rules that refer to defunct WAN policies have no action and the execution of rules continues until the next matching rule is found. Therefore, the order of WAN Rules is important for defining how traffic is handled when a policy becomes defunct.

Defunct processing allows an administrator to define a fallback behavior. For example, a WAN Rule can send traffic from the VoIP server to "Balance among all WANs with less than 100ms latency". If this policy becomes defunct because no WANs have less than 100ms latency, then a second rule can send traffic from the VoIP server to the "Best WAN" with the lowest latency.

Important: WAN Failover works best when the Policy Type uses Best WAN or Balance scenarios with specific WANs selected within those schemes.

Rules where the Policy Type uses Specific WAN do not failover unless the link is physically disconnected.

Have more questions? Submit a request

Comments

0 comments

Please sign in to leave a comment.