How Cloud Packet Core meets three top telco priorities
In my job I get to discuss the latest core network developments with my communications service provider (CSP) colleagues. It’s something I thoroughly enjoy. It’s also great to get their views on how the core network should help their businesses.
They take a pragmatic view. While the technology is fascinating, how it can help their organizations reach their goals is what really counts. There are three areas they often highlight – reducing energy consumption to save OPEX, improving network performance to deliver winning services and automation to optimize operational costs.
Let’s look at how Cloud Packet Core networks can help in each area.
Boosting energy efficiency
Dramatic energy price rises over the past year plus the ever-sharpening focus on sustainability is making energy efficiency more relevant every day. 5G core hardware and software that improves energy efficiency and reduces carbon emissions is high on the CSP shopping list.
Let’s take hardware first and look at four developments.
- Chipset performance: Using the latest 4th Gen Intel Xeon processors, Nokia has demonstrated a 30 percent performance gain in the 5G core User Plane Function (UPF) over the previous generation processor. This translates directly into a 30 percent generational improvement in energy efficiency. However, when you also factor in power management techniques, the gains reach as high as 43 percent run-time CPU power savings for substantial OPEX savings and emissions reductions.
- P-State and C-State capabilities: P-states enable the CPU frequency to be adjusted per core, while C-states allow the CPU to enter different sleep modes. The combination of both modes significantly reduces power consumption.
- Data Plane Development Kit, or DPDK: The latest library supports CPU cores entering C- and P-states directly from the application user space. In just one example, it uses a low-power sleep C-state while waiting for new network events.
- Kubernetes Power Controller profiles: This technology schedules Kubernetes nodes to switch between profiles – entering a high-performance profile during peak hours and moving to a power-saving profile during low traffic periods. This way, each application can request and use a specific power profile.
What about core network software? One major development here is the application of Artificial Intelligence (AI). With AI and Machine Learning (ML), Nokia’s Cloud Packet Core learns and predicts network patterns to improve processes such as network optimization, demand prediction and overload prevention, saving energy used by the processing units.
Strengthening network performance
Welcome to the world of integrated hardware-software appliances. Although not new, the technology is on the verge of much wider deployment. That’s because it is a flexible and cost-effective way to boost the performance of all shapes and sizes of network, whether based on a big, centralized cloud that serves densely populated areas, or a more geographically dispersed network with local optimizations that only appliances on the edge can deliver.
Appliances are pre-integrated in the factory to reduce deployment time and can be based on either two servers, with a user plane capacity up to 100 Gbit, or six servers, with capacity up to 340Gbit. They can be used on both the Control Plane and the User Plane to bring significant savings in footprint and power consumption. Their simplicity, supported by automation, is a further benefit – should more capacity be needed, an additional appliance can be easily and quickly deployed.
Network automation for higher operational efficiency
Automation is a powerful and fast-growing capability that promises to help CSPs optimize their operational costs. We believe CSPs need to adopt and adapt two operational practices proven in the hyperscaler world.
The first is to implement cloud-native design and orchestration. Kubernetes is a webscale orchestrator that can automate the lifecycle management of telco network functions, in the same way as it does for any webscale application. This enables the network to be expanded or shrunk as easily as any webscale application to precisely match demand. To do this means that the network must be considered as a collection of orchestrated and automated microservices rather than a jumble of disjointed network elements.
Secondly, CSPs can gain huge benefits taking the DevOps model and expanding it into DelOps which extends the Continuous Improvement/Continuous Delivery (CI/CD) approach from a hyperscaler’s environment to the telco world. This adaptation is needed because DevOps is inherently single vendor and doesn’t work in a CSP’s multivendor environment.
CSPs must address integration across multiple vendors while integrating CI/CD delivery streams from each vendor. Nokia’s DelOps delivery operations model addresses this challenge and brings true multivendor CI/CD agility to CSPs.
Although CSPs vary widely in terms of their scale and business strategy, I’ve learned they mostly want a choice of solutions to help them build and optimize their network architecture. Cloud Packet Core technologies such as those I’ve described here will continue to evolve rapidly to help them achieve their aims without constraints.