Algorithms for Network Functions Coordination and placement in Network Function Virtualization (NFV) simulated environment
Αλγόριθμοι για συντονισμό και τοποθέτηση λειτουργιών δικτύου σε προσομοιωμένο περιβάλλον Εικονικοποίησης δικτυακών λειτουργιών (NFV)
Master Thesis
Author
Κοψαχείλης, Χρήστος
Kopsacheilis, Christos
Date
2021View/ Open
Keywords
NFV ; VNF ; Algorithms ; Service function allocation ; Service function coordination ; Service function placementAbstract
Network Function Virtualization (NFV) is the current concept and the network architecture
used from the majority of the providers and operators that get rid of
network functions (such as firewalls, DNS, NATs, load balancers, intrusion detection
devices, WAN accelerators, etc.) from dedicated hardware devices. This decoupling
enables hosting of network services, known as Virtualized Network Functions
(VNFs), on commodity hardware (for example switches or servers) and thus facilitates
and speeds service deployment and management by providers, improves
flexibility, leads to more efficient and scalable resource allocation and usage, and
reduces in general the costs as the virtualized services can run on less expensive,
generic servers instead of proprietary hardware. This concept is a new chapter in
the evolution of networking, as it introduced high expectations for enhanced economical
network services, as well as major technical challenges that are currently
been researched.
This thesis addresses three baseline algorithms(Random Schedule, Shortest Path,
and Load Balance) along with the results achieved by applying them on the problem
of Coordination and Placement in a Network Function Virtualization (NFV) simulated
environment. That is the problem for coordination of service mesh consisting
of multiple microservices. This topic is always under constant analysis and research
from many operators , as the coordination of the services is a complicated a problem
and proposals for better solutions are currently analyzed from many Research and
University Departments.