Accelerating Service Time-to-Market through Network-driven Platform Customizations

CHALLENGES

As the adoption of 5G accelerates, Communication Service Providers (CSPs) find themselves in a mix between the vision and implementation. The vast set of use cases envisioned for different industries with 5G along with the need to transition from virtual to cloud-native networks are like a bright light at the end of the tunnel providing a sense of fast-paced, sustainable growth of technology over the next decade. Furthermore, the boom of AI and ML has increased the pace at which data is consumed, solidifying the case for technology innovation and the need for high speed networks. 

But these visions are coupled with implementation challenges for CSPs. One of the major challenges is managing the technical complexities introduced by 5G – revolving around containers and hybrid cloud deployments. These complexities pose support, workforce and financial problems to the CSPs. The implementation of infrastructure requirements of the network functions such as configuring SR-IOV interfaces, NUMA alignment, real-time kernel, etc. complicates the network operations. These requirements lead to numerous customizations of the underlying infrastructure platform that would require costly, time-consuming manual intervention and repetition across hundreds, possibly thousands, of instances, ultimately slowing down the service activation time.

THE SOLUTION

VMware Telco Cloud Platform simplifies the infrastructure customizations for the CSPs through dynamic infrastructure policies. The automation capabilities of VMware Telco Cloud Platform customize the infrastructure to meet the requirements of cloud-native network functions (CNFs) being deployed, ensuring the right state of the node during instantiation. This automated customization reduces the time and cost associated with infrastructure lifecycle management.

VMware Telco Cloud Platform adheres to the ETSI standards in designing and deploying the network functions, which are defined using TOSCA Simple Profile in YAML specification. A TOSCA template is a deployment description file that describes the deployment parameters of network functions and services on an infrastructure platform. It also lists the infrastructure requirements of the network functions. These templates are packaged in Cloud Service Archive (CSAR) format. 

VMware Telco Cloud Platform, through its automation capabilities, fetches the infrastructure requirements from CSAR packages and automatically customizes the infrastructure platform with those specifications. This ensures that the infrastructure is set up as required by the CNF, thus simplifying the lifecycle management of the infrastructure platform for platform engineers and administrators.

THE PROCESS

Let us walk through the instantiation of a network function to understand how VMware Telco Cloud Platform implements dynamic infrastructure policies. For our purpose, we will consider a sample CNF called “TestNF”. Before instantiating a network function, we have to onboard it – in other words – upload its CSAR file.

To onboard the TestNF network function, 

1. Log into VMware Telco Cloud Automation UI and navigate to Network Function under Catalog section. Click Onboard Network Function, provide a package name “testnf” and browse the CSAR file for TestNF. Then click Upload.
   
   

Once the TestNF network function is onboarded, let us examine the CSAR package to understand its infrastructure requirements – 

1. Navigate to Network Function under the Catalog section, and click testnf. Then navigate to the Infrastructure Requirements tab. 
2. Take note of the infrastructure requirements of TestNF – enabling I/O MMU, Hyper-threading, and real-time kernel, configuring NUMA alignment and VLAN interfaces, installing and setting up various linux packages, etc.
   
   
   
   
   
   

Let us verify the current state of the worker nodes in the workload cluster “wkld-01” where we intend to instantiate the TestNF – 

1. Log into one of the worker nodes in the workload cluster
2. Check the number of interfaces on the worker node – ip link
   
3. Check the hugepages details and usage – grep -i hugepages /proc/meminfo
   
4. Check the linux kernel version – uname -r
   
5. List the number of linux packages installed – sudo tdnf list –installed | wc -l
   
   

The next step is to instantiate the network function, to do so – 

1. Navigate to Network Function under the Catalog section and click the ellipsis next to testnf. Click Instantiate.
   
2. Furnish the name, click on the arrow in the Select Cloud section.
   
3. It opens a Select Cloud dialog. In that dialog, select the CaaS workload cluster ‘wkld-01’ as the VIM and click Next.
   
4. Select the node pool to deploy TestNF.
   
5. Review all customizations that will be applied to this node pool to prepare it for the deployment of TestNF and then click OK. Click Next.
   
6. In the Helm Charts tab, enter the namespace, repository URL, and any other required details, then click Next.
   
7. Click Next in the Network Function Properties tab.
8. In the Inputs tab, navigate to Testnf1 sub-tab. Provide the values file, select the portgroups for TestNF interfaces and click Next.
   
9. Review the network function instantiation details and click Instantiate.
10. To monitor infrastructure reconfiguration and network function instantiation, go to Inventory > Network Function and expand testnf.
    
11. After the infrastructure platform (specifically the CaaS cluster) is reconfigured and the network function is deployed, the state of the CNF testnf in Network Function Inventory will change to “Completed”.
    
    

Let us now validate that the infrastructure platform customizations have taken effect (validate at least a few of the customizations to compare with the pre-checks) – 

1. Log into one of the worker nodes of the specified node pool in the workload cluster
2. Validate that interfaces required for TestNF have been added on the worker node – ip link
   
3. Validate the hugepages configuration – grep -i hugepages /proc/meminfo
   
4. Validate that the real-time kernel is installed – uname -r
   
5. Validate that some linux packages have been installed – sudo tdnf list –installed | wc -l
   
   

This process demonstrates the use of dynamic infrastructure policies to automatically customize the infrastructure platform based on the specific needs of the CNF. With the dynamic infrastructure policies, VMware Telco Cloud Platform simplifies the lifecycle management of the infrastructure platform and network functions. The result prevents the overprovisioning of hardware resources and significantly shortens the time for pre-deployment configuration and validation, ultimately speeding up service time-to-market, reducing operational costs, and optimizing the network total cost of ownership.

To learn more about VMware Telco Cloud Platform, refer to the documentation here.