← Previous lessonFree lesson 3 of 3 · lesson 3 of 25

Unlock all 25 lessons of Telco CloudFull course ₹40,000 · paid

Price & syllabus
Free lesson · 19 min · no payment

Virtualized EPC, Cloud Native Functions and the 5G Core

A free lesson from Skill-Lync's Telco Cloud course, on Grow by Skill-Lync · Last updated 16 Nov 2026 · Session: “Week 1 Session 3”

What will you learn in this free Telco Cloud lesson?

The instructor explains why traditional purpose-built EPC hardware is rigid and how virtualized EPC uses generic hardware, NFVI and an orchestrator. The lesson contrasts VM-based and container-based network functions, lists EPC limitations, and maps MME, gateways and PCRF to AMF, SMF, UPF and PCF in the 5G core.

  • Problems with purpose-built EPC hardware
  • Softwareization, virtualization and orchestration in vEPC
  • VNF with VMs versus containers
  • EPC functions mapped to AMF, SMF, UPF, PCF

What comes after lesson 3?

Later in Skill-Lync's Telco Cloud course (paid): lesson 4, Week 2 Session 1; lesson 5, Week 2 Session 2; lesson 6, Week 2 Session 3. The full course has 25 lessons (12 h 32 min of video); see the course overview or its price and syllabus.

Lesson transcript

Machine transcript of the lesson audio (lightly cleaned). Technical terms may contain errors; the video is the reference.

When it comes to the benefits of virtualized EPC, the main benefit is provided by the use of generic hardware. Due to generic hardware, it can be reused. The same hardware that can be used for S-Gateway can also be used for virtualized E-NodeB. This means use of standard hardware for all network elements, easy to train the workforce, and simpler inventory management. Hence, reduce total cost of ownership. The network elements are not physical, rather virtualized here. So, increasing resources when there is more traffic or reducing the resources during off-peak hours is possible.

Applications are running on software, hence quicker to deploy and faster to upgrade. The instantization of new services are done in matter of minutes with the aid of pre-configured scripts. This means lesser time to market. The services can be rendered more quickly when compared to legacy hardware. Virtualization of services provides separation of network function and capacity. So, the capacity can be dynamically changed as per the network requirement. When the traffic increases, the capacity can be increased and when the traffic reduces, the capacity can be reduced. This is only possible thanks to the virtualized EPC.

When it comes to deployment, there are many challenges faced with traditional hardware. Traditional network elements are purpose-built hardware. Hardware, that means a specific network element, hardware can work as the network element only. Hardware built for serving gateway node, for example, cannot be used as some other node. Its software is also purpose-built to perform the functionality of the serving gateway. The hardware used will have specific configuration, which cannot usually be used in some other network element. For example, hardware card used in serving gateway will not be used in MME.

Due to dedicated hardware and software, the time of installation, configuration, and integration to system is higher. These processes are usually manual to great extent, hence our high service time to market. This kind of traditional deployment is quite rigid. That means in case of increase in traffic, the hardware cannot be readily increased. Next, new hardware needs to be installed, configured, and integrated in the system, which is costlier and time-consuming process. The legacy hardware is purpose-built and differs for different vendors. This requires skilled resources, various pairs to upkeep and to maintain the network, which increase the total cost of ownership of such traditional hardware.

To overcome the challenges experienced by CSPs using traditional hardware, many big telecom company operator came together to put an effort to come up with a solution. In 2012, they released first white paper to develop a new generic hardware. The list of CSPs that contributed can be found out on the screen. Virtualized EPC or Virtualized Evolved Packet Core works on three main principles. Softwareization, Virtualization, and Orchestration. In simple words, the idea of VEPC is to have a very capable generic hardware to be used a platform with use of softwareization create virtual network elements.

Or so-called network functions are created in it with help of various scripts. This is really interesting as physical network functions cannot be scaled up or scaled down. However, virtualized network functions can be increased or decreased in matter of minutes by running few scripts as underlying high capacity hardware is always available. Now, management of hardware like allocation, increasing resources or reducing resources can be done manually. But this will defeat the purpose of autoscaling of hardware as per the need before services gets affected. Here comes the role of orchestrator.

The orchestrator is always available manager who manages the resources and allocates hardware resources to the network function as and when required. When it comes to the benefits of virtualized EPC, the main benefit is provided by use of generic hardware. Due to generic hardware, it can be reused. The same hardware that can be used for S-Gateway can also be used for virtualized E-NodeB. This means use of standard hardware for all network element, easy to train the workforce and simpler inventory management. Hence, reduce total cost of ownership, the network elements are not physical, rather virtualized.

So, increased resources when there is more traffic or reducing the resources during off-peak hour is possible. Applications are running on software, hence quicker to deploy and faster upgrade. The instantization of new services are done in matter of minutes with the aid of pre-configured scripts. This means lesser time to market, the services can be rendered more quickly when compared to legacy hardware. Virtualization of services provides separation of network functions and capacity. So, the capacity can be dynamically changed as per the network requirements. Let us understand how the virtualization of network function is done.

The base layer is comprised of hardware resources. It includes compute resources, CPU, storage and top of the rack switches. It is stopped by a virtualization layer that will manage the underlying hardware resources and facilitate the sharing of hardware resources underneath. This virtualization layer offers virtual CPU, virtual storage and virtual network to the VMs on the top which are powered by actual hardware resources present underneath. These three layers, virtual resources, virtualization layer and hardware resources are part of network function infrastructure or NFVI. This is the virtualized infrastructure that is foundation over which multiple virtual network functions, for example, virtual S gateway, virtual P gateway, virtual MME can be created.

A virtual network function generally is made up of multiple virtual machines or VM where each VM performs a specific task. This whole operation is managed and supervised by network function virtualization manager and orchestration of NFV, MANU. Orchestration functionality manages the resources and provides them to VMs. When traffic increases, the VM needs more resources and orchestrator allocates the required resources from virtualized compute, storage and network. We discuss all these components in more detail as we progress in this course. The virtualization of network function can be done in two ways using VNF virtual network functions or cloud native network functions.

In VNF based deployment, the virtualization layer or hypervisor is managing the underlying hardware and provide the resources to the VMs. Please note that in each VM there is its own guest operating system, binary file and application that runs on the VM. The cloud native function based approach uses containers rather than VM. Containers allow users to package software, application, functions or microservices with all the files necessary to run it as it can be seen on the screen. Also, the operating system is shared between multiple applications. This approach makes it easy to move the contained component amongst environment for development, testing and production etc.

Now we will discuss about EPC limitations. Since beginning, the motive of technology has evolved over time. First generation technology was introduced to introduce mobility to otherwise static telecommunication services. Later, due to poor quality and non-availability of roaming, second generation technology GSM came into existence. When we are in the open cell phone, it's a real challenge that allows us to pick up with the delictory grid of the river and the square foot of the river and the river. The other generation technology was a bit less than 1,000th generation L-R, that was George Martin Humbly and the other generation.

The other generation of devices used in the network of the metro and the public extension of the river and the river is now, that we will also assemble, and finally start with the river and the river to help us with the center port of the river. After it loads from the river and the river between the river to river and the river, we are near the river to the river and the river but the river to the river to the river and the river. It initially introduced circuit switching data and later packet switching connectivity with GPRS and Edge.

Around 1999, the focus of CSPs was on packet data as much it was on the voice services. Hence, the UMTS supported voice services and decent throughput for data. The fourth generation technology, LTE, is designed to support very high data rates compared to predecessors. LTE is used to transport huge volume of data packets to and from the UE. Later, with introduction of Internet of Things, the LTE is customized to support machine-to-machine communication using NB-IoT, narrowband IoT, or LTE-M. But this is not considered to be the efficient mode of transport as LTE is designed to provide high bandwidth to the users and generally, IoT devices produce small amount of user plane data.

Since the future is going to be more about device-to-device connectivity as much it is going to be for human, a better design technology is required to support extremely high data rates, greater than 10 Gbps for user and support machine-to-machine communication, demanding very low latency in the order of around 1 ms. This cannot be fulfilled using Evolve Packet Core and new improved 5G core network is required to meet these requirements. In EPC, there are also some other limitations. EPC uses too many protocols, it has scalability issues, the network slicing is there, but it is not completely supported as it should be in case of 5G core network.

Its architecture is more rigid as compared to the new 5G core architecture. The architecture of EPC is mostly centrally located, that means all the network functions are placed at some common central location, which generally prevents achieving a low user plane latency, which is required now. Now we will discuss evolution towards 5G core network. As EPC LTE failed to support all the use cases seeking large control plane signaling and ultra-low latency, the new designed 5G core network is required to meet the increasing demand of user plane data.

4G targeted 1 Gbps of downlink throughput, whereas 5G supports up to 10 Gbps and beyond. 5G core will be deployed in virtualized environment, hence easily scalable. Massive machine-to-machine communication or MMTC use case will demand huge control plane resources to meet this demand from large number of IoT devices. Distributed core architecture of 5G along with improved radio network, new radio will support for ultra-low latency, which is less than 1 ms required for ultra-reliable low latency communication use cases. For example, discrete automation in factories for autonomous driving cars, etc.

Thus, to support all the use cases, extreme mobile broadband, massive machine-to-machine communication, and ultra-liable low latency communication, a better core architecture is required, which is provided by 5G core network. There are many salient features of 5G core network. It is a software-driven network and IT. Network functions are virtualized and applications are cloud-native, instantiated using the container services. Thanks to distributed core architecture, the placement of various network functions is flexible. The network resources can be centralized or present at the far edge very close to the user, based upon the latency requirement.

The 5G core network is access agnostic. It can be connected to 5G new radio, or LTE-EUT-RAN, or a non-3GPP technology as Wi-Fi. The authentication of users from various radio access technologies is done using common authorization node. Unlike EPC, where the session state of UE, like UE context, service information is only stored in the network elements. In 5GC, the session state may be stored in common storage space, which can be accessed by other network elements too. 5G core supports network slice completely ensuring complete isolation of services for different quality of service, type, or even for different customer based upon the service level agreement or SLA agreed with them.

The 5G core network is overhauled core architecture to meet the modern requirements. However, functionality of many network functions is inspired from EPC. For example, MME functionality is now network function, referred as access and mobility management function, or AMF. Similarly, user plane functionality of S-Gateway or P-Gateway is user plane function, or UPF. PCRF of EPC is now policy control function, or PCF. Of course, there are many new network functions like NSSF, NRF, UDSF that had no counterpart in EPC. We will learn about these network functions in detail as we progress in this course.

The initial EPC network followed traditional box kind of approach, where a network element is separate unit which performs various functions. For example, as you may see, MME, for instance, has two functional elements. Control plane that manages the non-access stratum or NAS signaling of the UE, and session state that stores the UE context. Similarly, serving gateway and PDN gateway has control plane functionality for bearer management and handling and the user plane block that is responsible for forwarding and routing of user data packets. The functional block for session state hold the session information of the active UE.

This makes bifurcation of control plane from user plane impossible. With this kind of rigid architecture, the control plane and user plane handling function have to be at the same location. In 5GC, the network elements are decomposed into network functions. The network functions are defined as 3GPP adopted or 3GPP defined processing function in a network which has defined functional behavior and 3GPP defined interfaces. The network function in 5G performs only a specific function unlike EPC where a network element performs multiple functions as discussed in previous slides.

For example, a S-gateway in EPC manages control plane and does packet routing also. In 5GC, these functions are separate. Control plane handling goes to control plane referred as session management function or SMF. And routing forwarding of user plane packet is done by user plane function or UPF. We will discuss in detail about SMF and UPF in upcoming modules later. But at this time, it is important to understand that in 5GC, the architecture is modular where each module performs a specific functionality. In this session, we learnt about the architectures of 2G and 3G, the definition and need of LTE EPC offers, the evolved packet system, and the need of CUPS, the generic hardware and need of virtual EPC, the introduction and importance of 5G core.

For image credits, thank you.

Want the rest of Telco Cloud?

Get the full course syllabus and fees.

The full course is paid (₹40,000). We'll send the syllabus and call you back within one working day.

Talk to a career counsellor

Free 20-minute call. We'll map the right course to where you want to work.

By submitting you agree to our Terms and Privacy Policy.

Full course · paid

Unlock the full Telco Cloud course

The remaining 22 lessons, mentor-reviewed projects and a certificate come with the paid course (₹40,000). Leave your details: we'll send the syllabus and fees and call you back.

Get the syllabus and fees

A counsellor calls you back within one working day.

By submitting you agree to our Terms and Privacy Policy.

Grow by Skill-Lync · free lesson

Start watching

Free, no payment. Leave your details once and every free lesson on Grow plays straight away. A Skill-Lync counsellor may call you about the full Telco Cloud course.

By submitting you agree to our Terms and Privacy Policy.

Talk to a career counsellor

Talk to a career counsellor

Free 20-minute call. We'll map the right course to where you want to work.

By submitting you agree to our Terms and Privacy Policy.