10. Skip to content

10. Traffic Policy

Traffic Policy (aka Traffic Steering, Traffic Engineering) is a set of features that allow you to determine how certain classes of traffic should be forwarded from your sites to the other sites or to the general Internet.

It combines traffic classification, filters, policy-based routing, and QoS in a user-friendly manner.

This section describes some of the specific use cases that could be configured using this feature.

All traffic steering is applied in only one direction, from the LAN ports to the WAN ports.

There is only one feature that works on traffic from the WAN ports to the LAN ports, which is DNAT or Dynamic Network Address Translation, and it is described in a different section.

10.1 LAN Traffic Filtering

If traffic from certain LAN segments must be discarded, specify the LAN port, IP subnet information or 5-tuple information and select 'DENY' as the action item.

In many cases, it is essential to discard traffic from some LAN ports so that they are prevented from accessing the Internet or even the enterprise traffic.

A very common use case is to specify a list of social network domains and discard those traffic. That way, your enterprise premises is not used for social network apps.

ChiefNET allows you identify specific traffic flows based on any of the following attributes in a flexible manner.

  • Source IP/subnet, Lists
  • Destination IP/subnet, Lists. List of Domain Names, Groups of Domain Names
  • Protocol
  • UDP/TCP Source Port, Lists, Ranges
  • UDP/TCP Destination Port, Lists, Ranges

The above parameters completely encompass all ways of identifying a specific set of traffic flows, that can be tied to applications.

10.2 Application Traffic Filtering

Nowadays, applications are identified by domains. A malicious web application is typically avoided by entirely turning the domain off. A domain is under the administrative control of an organization. A single domain may result in multiple IP addresses/multiple hosts. Use the traffic steering feature to specify a list of domains or domain-groups to discard traffic to them, or to configure a specific quality of service to them.

WHen you need URL filtering, use domain-filtering. URL filtering is a means to avoid specific sites within a www domain. It was a feature mainly used when a single proxy is used to access multiple WWW domains, by attaching one URL suffix to each WWW domain/virtual host. Hence to turn off a single WWW domain/virtual host that supplies one application, URL filters were employed in the past.

From a security point of view however, it is best to avoid the entire domain, as it would be natural to assume that if one application is compromised, the others under the control of the reverse proxy are also under the same vulnerability.

For this reason, we highly recommend the use of domain filters instead of URL filters. While URL filters can satisfy short term needs, domain filters are more effective at better security.

Where URL filters are required in your own enterprise, we highly recommend the use of alternate solutions wherein the proxies employ the appropriate security mechanisms (e.g. SSO) to ensure that malicious users or malicious services are prevented in your network.

10.3 Enterprise-Only LAN or Systems

There may be machines on your LAN premises that must only be connected to the Enterprise network and not directly to the Internet.

In this case, specify the machine's IP or the entire IP subnet, and select 'ACCEPT' as the action. You will then have to choose the Enterprise network interface (i.e. the VPN tunnel interface). This is helpful when specific remote systems must only be managed from the enterprise network using secure transport methods.

This is typically the case with certain video surveillance camera systems, point-of-sale terminals, printers etc. that do not need access directly to the Internet.

10.4 Quality of Service

For certain VOIP traffic, or connections to certain domains such as Microsoft 365, you might want to specify a minimum bandwidth and a maximum bandwidth. These may first be specified as a QoS profile and the QoS profile may then be used in the Traffic Steering section.

10.5 Application-Specific Traffic Failover and Load Balancing

Part of QoS is the ability to be resilient when the network fails on the ISP side. In this case, you may configure the primary WAN port and the secondary WAN port. You can also configure the specific criteria such as the latency, jitter or packet loss thresholds to initiate a switchover to a better-performing WAN port.

10.6 Exclusive WAN Port Selection

For certain classes of traffic you can choose a specific WAN port to be used, and only that WAN port. A common use case is when you have a backup WAN port that should not be overwhelmed with low-priority or unimportant traffic when the primary WAN port fails. This is typically the case when the backup WAN port is expensive for the amount of bandwidth and the data allowance.

10.7 Default Traffic Failover

If there are multiple WAN ports, ChiefNET continuously monitors them for Internet access and switches over to the working WAN port. This ensures that there is at least one Internet-access port even if no traffic steering rules are configured.

10.8 Additional Notes

There are many use cases as mentioned above for the integrated traffic steering feature. We highly recommend that all of the features be documented ahead of time during the proof-of-concept phase to ensure that the implementation meets the requirements.

10.9 Configuring a Traffic Filter to Discard

Configuring a traffic filter implies that there is a class of traffic from the LAN ports that should be discarded, without getting forwarded to the WAN side.

This is done on the traffic steering section

  • Select the interface from which the traffic originates
  • Specify the 5-tuple information such as source subnet, destination subnet, IP protocol, source port range and the destination port range. Not all information is mandatory. The traffic classification can be as specific as necessary.
  • Select DROP from the 'action' drop down menu.

When the selected traffic is 'dropped', the configuration portion related to failover, load-balancing, or QoS will be disabled.

10.10 Configuring WAN Port Load Balancing

When there are two ISPs on a given CPE, and there is a need to utilize both the ISPs, one on each WAN port, the CPE device can be configured with 'load balancing' for any classification of traffic.

  • Configuring the 5-tuple parameters that identifies the class of traffic
  • Select "ACCEPT' as the action
  • The traffic policy portion of the configuration screen will be enabled. Here select the load-balancing tab, so that the selected class of traffic is load-balanced.

It is possible to configure multiple such classification rules, and select the same 'load balancing' configuration.

The CPE figures out multiple microflows within the selected class and automatically chooses the WAN port that is currently more available. All load-balancing is done at the microflow level, and not at the packet level to avoid unnecessary packet out-of-order/reorder issues as those will hamper all the applications in an enterprise network.

10.11 Configuring Primary/Backup WAN Ports

Similar to the above, for a traffic class, it is possible to configure the primary WAN port and the secondary WAN port.

The configuration is very flexible. For some classes, the primary could be WAN-1 and secondary could be WAN-2. For some other classes, the primary could be WAN-2 and the secondary could be WAN-1.

It is even more flexible in another aspect. It is possible to configure an exclusive WAN port so that the traffic stream is discarded if that WAN port is not available. THe main use case here is that for non-important traffic, the expensive or low-bandwidth backup links should not be used. In most cases, the backup links should be used only for highly important traffic. By configuring the traffic streams like so, it is possible to ensure that the customer's critical traffic streams have primary+backup available at all times.

10.12 QoS/SLA Based Traffic Steering

It is possible to configure quality-based selection of the WAN port for a given trafic class.

For example, there may be a class of traffic that requires a low latency WAN port if available. In order to do this, the required latency may be configured in the Primary/Backup configuration tab. The CPE software automatically put the traffic on that class on the WAN port that is the most suitable, given its latency/jitter/loss parameters.

A backup WAN port may still be assigned and that will be used in case the primary WAN port is not available.

This feature is very unique to ChiefNET and is useful for specialized applications. The main use-case for this is that for some class of traffic, the preference should be given to some WAN links that have a specific SLA obtained from the ISP.