• Hyperconverged Infrastructure (HCI)
• Kubernetes
• Proxmox VE (PVE)

The purpose of my thesis is to characterize end-to-end latency of an application installed in a
hyperconverged structure in a standard and critical scenario which consists in a Load, stress
test and both tests at the same time.
In last decades the telecommunication sector is growing rapidly caused by an exponential increase
of subscribers which exceeded eight billion for the first time in 2019 and arriving at 8.9 billion mobile
subscriptions worldwide in 2023 with a global population of 7.9 billion as of 2022. This means which
the number of smartphone subscriptions now significantly exceeds the number of people on the
Planet.
This expansion of subscribers means more bandwidth demand, scalability, flexibility, agility and
security.
All these requirements were virtually impossible to manage with traditional network architectures
because network designers are constrained by the limitations of traditional network:
- Complex and custom devices: Protocols tended to be defined in isolation with each solving a
specific problem, customizing the problem, and without the benefit of any fundamental
abstractions. Devices required specific configurations which is not easy and quick in large
networks while there is absence of global vision of the network.
- Redundancy and over provisioning: traditional network architecture was designed to ensure
reliability and to improve fault tolerance on special occasions, such as peak traffic times or in
case of a hardware or software failure.
- Inability to scale because the traditional networks architecture is static.
Therefore, the telecommunications sector has developed new technology and new network
architectures like virtualization, Software Define Network and Network Functions Virtualization.
These new paradigms, in addition to guarantee the new networks requirements, have allowed
network industry to provide new benefits like latency reduction, bandwidth increase and even more
reliability which are managed by the last mobile network generations but mostly in 5G.
5G offers a range of specific benefits over previous iterations:
- Enhanced Mobile Broadband (eMBB).
- Massive Machine Type Communications (mMTC).
- Ultra-Reliable and Low Latency Communications (URLLC).
These new benefits have made telecom companies need to keep pace with challenges like IoT, Gen
AI, cloud computing, V2X and many more.
To offer these services 5G makes use of a set dedicated technologies like End-to-End (E2E) Network
Slicing and Multi-Access Edge Computing (MEC).
As more of devices becomes connected to new 5G services the internet protocol (IP) traffic will
skyrocket, increasing the utilization of data centers which see most of the world traffic and big data
pass through them.
In 2018 people used 26.7 EB per month mobile data in total, Figure 2, where one EB is equivalent to
one billion GB. And in according to the estimates, the amount of data used on mobile devices is
projected to increase significantly, reaching a staggering 483 EB per month by 2028.
Therefore, data centers architecture must be involved to manage traffic increase.
Data center architecture evolutions passed from traditional to converged to finish in
hyperconverged infrastructure which is a software-centric architecture that tightly integrates
compute, storage and virtualization resources into a single, highly integrated system, along with a
comprehensive software stack that includes operating systems, virtualization platforms and
software management tools.
This new architecture provides several advantages which will be analysis in this thesis.
Can a hyperconverged infrastructure integrate 5G services, such uRLLC?
Hyperconverged architecture is based on the automation of infrastructure, allowing easer
installation of physical machines and applications enabling virtual machine migrating from a server
to another in critical event. This automation leads hyperconverged solutions as a candidate to future
mobile scenarios beyond 5G based on Edge Computing.
Typically, such mobile radio scenarios will apply to URLLC services which require to be always
available, 5 nine rule which allows approximately 5 minutes of downtime per year, and to have a low
and controlled e2e delay. A current example of these services is industry 4.0, which involves
remotely controlling the entire production chain of a factory.
The hyperconverged architecture using automation is able to meet availability requirements but
introduces multiple levels of virtualization which introduce additional delays into the application
process.