Carrier Ethernet networks today are migrating from being uni-service network (i.e., best effort) to being a multi-service network (prioritized). In order to run a multi-service network the carrier needs a mechanism that allows him to enforce QoS (Quality of Service) policies per user per service.
HQoS (Hierarchical Quality of Service) allows the carrier to make sure each service gets the network resources it needs while coexisting on the same network infrastructure with other types of services. An additional value of HQoS is the ability to share unused network resources between different services. This optimizes network resources while keeping the same level of service and eliminates allocating bandwidth to unused services.
What is HQoS?
HQoS is a tiered quality of service technology that classifies and schedules traffic based on service type and/or COS (Class of Service).
HQoS allows a carrier to consolidate different services on the same physical box on the same physical infrastructure. This is mainly done by the ability to have a policy per service, or even a policy inside a service. We can look at different packet types and different flows and give them the correct priority, the correct bandwidth requirements and the correct policy requirements.
Classical QoS examines the packet puts it into a queue and runs a scheduling method – strict priority, weighted round robin or other technology. It is limited to 8 queues and 1 level of scheduling and priority.
HQoS on the other hand can have 2 to 4 levels of scheduling and prioritization This means that you can look at each service individually for each customer. While you may have only one big pipe for services, you can give the customer smaller pipes each one for a different application, each one with its own prioritization requirements, each one with its own quality of service policy.
Why is HQoS needed?
In a today’s network, carriers need to run bandwidth hungry application like video services, alongside time sensitive applications voice over IP services or wireless backhaul traffic. Each of these services has its own set of requirements.
Wireless backhaul is very sensitive to fluctuations in the packet stream because it relies on synchronization and it’s a voice-based application. The same is true for voice over IP. Delay in a voice over IP environment would basically kill the service. On the other hand, video applications are bursty and delay doesn’t matter if the traffic is buffered. But from a bandwidth perspective, it is very demanding. Any congestion in the network will prevent a video from running.
If you add business customers on the same infrastructure, each business customer has their own SLA and their own requirements – some run different services – some combine voice applications with storage applications, with critical data applications like bank transactions that require specific quality of service demands.
For example, HQoS is implemented on our T-Marc 300 at the demarcation level. It can serve up to 10 customers and each customer can run up to 64 different services. So you can imagine for each port a different customer can be connected, and underneath this port, virtual lines would run different services, and each service can actually receive its own policy regardless of the same service in a different customer. So rather than having a single policy for voice, one for video, one for critical data, you can define a policy per customer per service rather than relying on a single policy per service globally.
What’s the easiest way to implement HQoS?
We visualize our devices as service machines. This means that the device transports services across the network. Using our service management system the EdgeGenie the user only needs to engineer a few general HQoS service package policies which can then be re-used on all other devices. Whenever a new service is added the technician only needs to apply the HQoS policy name and participating customers in the service and the policy is implemented across the path. This saves time and eliminates mistakes that might occur for each new service that is being added.
How can HQoS help a cellular wireless operator support more services?
Cell towers are sometimes located in good strategic places like industrial parks or city centers, where the operator might want to start offering business services alongside with its cell tower traffic.
One of the main challenges in such an installation is how to make sure that the sensitive cell traffic will not be affected by the new business services running on the same infrastructure. In this example, each type of traffic has its own subsets of internal priority flows and bandwidth constraints. It is exactly in these types of applications where HQoS can support such a mixture of services, allowing each of them to get the required resources and quality.