Why WAN optimization needs QoS
Without QoS, bandwidth-demanding applications fill the bandwidth that WAN optimization frees up. Initially this might mask the problem, but as demand increases it eventually resurfaces. WAN optimization also changes traffic signatures, much like encryption does, so traffic must be classified before it is optimized.
QoS resolves this by applying priorities to specific traffic, shaping certain traffic, or dropping unwanted traffic. The high priority class is designed for low-latency traffic such as voice. Most traffic flows sit in the normal priority queue, and the low priority queue is for best-effort traffic such as email. Bandwidth shaping can be applied where specific traffic starves other flows of network resources.
How it works
- Classification: a traffic policy rule identifies the interesting traffic, for example all traffic from a voice subnet, or SMTP on TCP port 25. Rules can be grouped in a class, so the class shares one action.
- Marking (optional): DSCP values mark traffic, which routers then use to shape it or to calculate the drop probability. For example voice is marked EF and SMTP AF13.
- Action: shape (limit the total bandwidth), drop (filter unwanted traffic destined for the WAN), or high, normal or low priority.
Layer 7 applications and firewall
Wanos identifies over 200 Layer 7 and 7,000 Layer 4 applications. Layer 7 QoS and visibility let the administrator identify, control and prioritize the top-talking applications, and the Layer 7 firewall filters traffic on deep packet inspection application signatures at the branch edge.