Foundry NetIron M2404C and M2404F Metro Access Switches
Configuring HQoS (Rev.03)
Carrier Grade Ethernet Services and HQoS
© 2008 Foundry Networks, Inc.
Page 9 of 98
Hierarchical QoS Usage Examples
Some examples of the possible HQoS model implementations are described below:
Example 1: Multi-Application SLA
In this example a customer has several applications. The customer has a single site and the packets
streamed from/to this site are classified into the various applications by EVCs (VLANs) over the
same UNI (physical port), or VLAN priority values (VPT) over the same VLAN.
Legacy QoS model implementation:
In this implementation each application would be limited by its own separate SLA, and if a high-
priority application would not use all of its CIR bandwidth, the unused bandwidth would be
wasted. This would also require the customer to purchase more bandwidth that he really requires,
since the bandwidth can’t be shared between the applications.
HQoS model implementation:
Each application receives its own set of SLA attributes (e.g. CIR - reserved bandwidth), but
another set of SLA attributes is defined on the customer level. Lower priority applications (e.g.
best effort internet traffic) can burst to use all the bandwidth available to the customer when the
higher-priority applications don’t use this bandwidth.
Unified
SLA
Voice
Video
Business
Best Effort
Figure 4: Multi-Application SLA
Example 2: Multi-Office SLA
In this example a customer has two offices, which are both part of the same site (e.g. located in the
same building). The offices connect to same VPN and share a single SLA. Each is connected to a
different UNI or a different EVC over the same UNI. Each office has its own CIR, but can burst to
use any available bandwidth. For example, if each office has CIR = 5Mb/s, and the total PIR is
10Mb/s, then if office #1 currently uses 2Mb/s only, office #2 can increase its traffic up to 8Mb/s
without losing packets.