As communications service providers adopt cloud and virtualization for business operations, they will disaggregate their networks. In this article, we look at why a new resource management model is needed for network disaggregation to work well and what structure that model should have.
A New Resource Management Model for Communication Networks
An Evolution of Networks: Network Disaggregation
By Xinchen Gao and Massimo Fatato
Over the last few years, communications service providers have considered cloud and virtualization as ways to gain flexibility and agility for their competitive service portfolio. The core principle of this new approach is network disaggregation, i.e. the separation between hardware and the software that supports network functions.
Historically, network devices - such as routers and switches - were sold as appliances, with the hardware and software bundled together. This has left little room for network operators to customize their solutions as they had to buy the devices as a single package from the vendor. As the network industry moves towards virtualization, network disaggregation has become a possibility. With network disaggregation, the individual hardware and software components of each device can be purchased separately, even from different vendors. This means that network operators now have scope to build their own custom solutions, optimized to fit their specific needs.
Network aggregations changes the network system in the following ways:
Even though network disaggregation benefits the service providers in many aspects, it also brings challenges, particularly in the way network and services are managed. Service providers require a resource management model which integrates all the sub-services together in an effective way and - at the same time - is highly flexible and can easily adapt to their network infrastructure.
A Need of Suitable Resource Management Model
To enable the integration, flexibility, and the adaptability that network disaggregation needs to work well, an appropriate resource management model would be needed.
An ideal resource management model has a layered architecture, with discrete components responsible for the needs of each layer. Where there are changes that impact functionality in a specific layer, resolution will be attempted internally within the same layer. If this is not possible, the service requested by the layer above is impacted and the components will report failure up the line.
Using SDN as an example, the model will act in two ways when there is a network failure:
Re-routes the packet stream to avoid impact on service availability (self-healing)
Flags any resource problem which cannot be re-routed or self-healed to the Service Orchestrator to take further action.
An Example of Resource Management Model Architecture
Within the resource management model, services are composed of multiple sub-services (or service components). A service component is built on resources that often associated in resources pool to guarantee the maximum availability. The resource pool is made of infrastructure components (e.g. a group of devices) as illustrated in figure 1. The model is ideal for network disaggregation because of its layered architecture.
The Benefits of Resource Management Model
Today, network and service information are kept in an inventory management systems that require costly reconciliation activities to maintain alignment with the live network. In contrast, within the resource management model, network status is carried by each sub-service; hence avoiding the need for inventory reconciliations. The benefits of introducing a resource management model are shown below:
Network disaggregation is already underway. Service providers that adopt a resource management model will overcome the challenges of traditional approaches to inventory management. <>
Cartesian has developed a network approach called ‘the Fractal Model’, which incorporates the resource management needs of effective network disaggregation. Please contact us for more details.