The objectives and benefits to be achieved in this research are:
- Understand how a microservice-based SDN Controller works.
- Understand how Network Slicing works.
- Able to implement an effective microservice-based SDN Controller system for Network Slicing.
- Able to know the effect of Microservice-based SDN Controller on Network Slicing performance on Fat Tree Topology.
In this research, Ryu SDN Controller based on microservices used as a multi-tenant control of the system. Every sub-function in the Ryu SDN Controller that is still monolithic will be broken down and each of these subsystems will be run using a Docker container and managed using Kubernetes. Then as a controller and slice maker on the data plane, the Flow Space feature from the Flowvisor Controller container-based will be used to enable some isolation from each tenant.
The fattree topology uses a slicing scenario for several hosts consisting of services commonly used by end-users and has a bandwidth consumption allocation that is not too large, including the following:
| Service Type | Slice Name | Bandwidth | Slicing Path |
|---|---|---|---|
| Video Streaming | Left Slice | 3 Mbps | H1-H4 |
| Web Browsing | Center Slice | 1 Mbps | H5-H8 |
| VoIP | Right Slice | 500 Kbps | H9-H12 |
In this research the research was conducted using 4 Virtual machines (VM) which it is built on top of Openstack.
| VM | Image | Spec | IP |
|---|---|---|---|
| Master | Ubuntu 20.04 | 4 vCPU, 4GB RAM, Storage 30GB | 10.0.0.241, 172.20.3.237 |
| Worker 1 | Ubuntu 20.04 | 8 vCPU, 4GB RAM, Storage 50GB | 10.0.2.207, 172.20.3.178 |
| Worker 2 | Ubuntu 20.04 | 4 vCPU, 4GB RAM, Storage 30GB | 10.0.0.128, 172.20.3.242 |
| Flowvisor | Ubuntu 14.04 | 2 vCPU, 2GB RAM, Storage 20GB | 10.0.1.242, 172.20.0.119 |