Alfio Lomb Alfio Lombard ardo o Anto tonio Manzalini Vincenzo - PowerPoint PPT Presentation

Alfio Lomb Alfio Lombard ardo o Anto tonio Manzalini Vincenzo Vincenzo Ricco iccobene ene Telecom Italia Gio Giova vanni nni Sc Sche hemb mbra Strategy Future Centre DIEEI – University of Catania

� Pa Paper er moti tivati tion an and d ref referen erence ce s scen cenario ario � Netw twork analyti tical fr framewo mework ◦ Model of an NFV node ◦ Model of a non-NFV node ◦ Model of the whole network ◦ Derivation of performance parameters � Case Case stu tudy � Conclusions Conclusions and futu ture work

� Service Providers and Netw twork Operato tors ne need: ◦ Flexibility in network deployment and management ◦ A flexible and optimal provisioning of network functions and services could reduce equipment costs and allow to postpone network investments ◦ New network functionalities, services and policies to increase dynamicity of the market ◦ Reducing OPEX and CAPEX

� SDN DN: Softw tware De Defined Netw tworks ◦ Decoupling the software control plane from the hardware data plane (packets forwarding), and moving its logic to centralized controllers � NFV: Netw twork Functi tion Virtu tualizati tion ◦ Virtualization of some network functions that can run on standard HW, and that can be moved and instantiated in various locations of the network

Current Current approach approach ≡

NFV NFV approach approach Virtual machines ≡ General purpose server Data center

NFV NFV approach approach Virtual machines ≡ General purpose server Data center

Virtual machines � An An “NFV “NFV no node” ” is is characte terized by by: : ◦ A standard hardware architecture (x86 commodity hardware) ◦ A virtualization capable software architecture ◦ A set of Virtual Machines (VMs) that run Network Functions (e.g. Routers, Firewalls, Load Balancer, ...) VM VM VM 2 3 1 Software Layer Hardware Layer

� Analy Analysis sis of th the impact t of th the Netw twork Functi tion allocati tion � An An analyti tical fr framewo mework fo for 
 perf perform orman ance ce evaluati tion of of 
 th the netw twork

E2e path for each flow Routing Analytical Protocol Model ◦ Network topology ◦ Performance parameters ◦ Network Function allocation ◦ Traffic characterization

� Let t us consider th the netw twork represente ted by a directe ted graph G ( V, E ), whe where: ◦ V is a set of vertices ◦ E is a set of links among them � Let t F be th the set t of functi tions deployed over th the netw twork

� User tr traffic is represente ted by a set t S of of f flow lows, , each characte terized by th the following ite tems: ◦ ∈ V is the vertex that represents the source of σ s the flow s ◦ ∈ V is the vertex that represents the destination δ s of the flow s ◦ f s is the mean bit rate characterizing the flow s ◦ func s is the set of functions required by the flow s

NFV node ( OUT ) Λ ( OUT ) Q ( F ) i , 1 i , 1 ( F ) Λ Q non-NFV node i , 1 i , 1 ( OUT ) ( OUT ) Q Λ ( F ) i , 2 i , 2 ( F ) ( OUT ) Λ Q Λ ( OUT ) Q i , 2 i , 2 i , 1 i , 1 CPU ( OUT ) ( OUT ) Q ( OUT ) ( F ) Λ ( F ) ( OUT ) Q Q Λ Λ i , 2 i , 2 i , j i , j i , h i , h ( OUT ) ( F ) ( F ) Q Λ Λ ( OUT ) ( OUT ) Q ( OUT ) ( F ) ( OUT ) Q Λ i , L i , L ( F ) i , L i i ( OUT ) i , h i , L i , h i i ( OUT ) ( OUT ) Q Λ ( OUT ) i , L ( OUT ) i , L i i

An NFV node can be modeled as a set of queues, that belong to two categories: NFV node Q ( F ) • Functi tions Queue Queue i , j ( OUT ) Λ ( OUT ) Q � They manage the access to ( F ) i , 1 i , 1 ( F ) Λ Q i , 1 i , 1 the functions ( OUT ) � Their service rate depends ( OUT ) Q Λ ( F ) i , 2 i , 2 ( F ) Λ Q on the CPU processing i , 2 i , 2 CPU s p e e d t o p r o c e s s t h e relative function ( OUT ) ( F ) ( F ) ( OUT ) Q Q Λ Λ i , j i , j i , h i , h • Outp tput t qu queu eues es Q ( OUT ) i , h ( OUT ) ( F ) ( F ) � They manage the packet Q Λ Λ ( OUT ) Q ( F ) ( OUT ) i , L i , L ( F ) i , L i i ( OUT ) transmission on the output i , L i i links � Their service rate depends on the output bitrate

Functi tion Queues Queues NFV node ( F ) ∑ Arrival Λ = λ i , j k Rate ( OUT ) k ∀ ∈ Φ Λ ( OUT ) Q i , j ( F ) i , 1 i , 1 ( F ) Λ Q i , 1 i , 1 ( F ) p ⋅ C ( CPU ) Service µ = ( OUT ) ( OUT ) Q i , j i , j i Λ ( F ) Rate i , 2 i , 2 ( F ) Λ Q i , 2 i , 2 CPU Outp tput t Queues Queues ( OUT ) ( F ) ( F ) ( OUT ) Q Q Λ Λ i , j i , j i , h i , h Arrival ( OUT ) ∑ Λ = λ i , h k Rate k ( OUT ) ∀ ∈ Ψ ( F ) i , h ( F ) Q Λ Λ ( OUT ) Q ( F ) ( OUT ) i , L i , L ( F ) i , L i i ( OUT ) i , L Service i i ( OUT ) C ( NIC ) µ = Rate i , h i , h

node i and requiring the function j : set of flows routed through the Φ i , j Functi tion Queues Queues NFV node Arrival ( F ) ∑ Λ = λ Rate i , j k ( OUT ) k ∀ ∈ Φ Λ ( OUT ) Q i , j ( F ) i , 1 i , 1 ( F ) Λ Q Service i , 1 i , 1 ( F ) p ⋅ C ( CPU ) µ = Rate ( OUT ) ( OUT ) Q i , j i , j i Λ ( F ) i , 2 i , 2 ( F ) Λ Q i , 2 i , 2 CPU Outp tput t Queues Queues ( OUT ) ( F ) ( F ) ( OUT ) Q Q Λ Λ i , j i , j i , h Arrival i , h ( OUT ) ∑ Λ = λ Rate i , h k k ( OUT ) ∀ ∈ Ψ ( F ) i , h ( F ) Q Λ Λ ( OUT ) Q ( F ) ( OUT ) i , L i , L ( F ) i , L i i ( OUT ) Service i , L i i ( OUT ) C ( NIC ) µ = Rate i , h i , h

p , assigned to VM (function) j : the CPU quota of i-th node i j Functi tion Queues Queues NFV node Arrival ( F ) ∑ Λ = λ Rate i , j k ( OUT ) k ∀ ∈ Φ Λ ( OUT ) Q i , j ( F ) i , 1 i , 1 ( F ) Λ Q Service i , 1 i , 1 ( F ) p ⋅ C ( CPU ) µ = Rate ( OUT ) ( OUT ) Q i , j i , j i Λ ( F ) i , 2 i , 2 ( F ) Λ Q i , 2 i , 2 CPU Outp tput t Queues Queues ( OUT ) ( F ) ( F ) ( OUT ) Q Q Λ Λ : the mean packet processing rate C ( CPU ) i , j i , j i , h Arrival i , h i of the processor in the i -th NFV node ( OUT ) ∑ Λ = λ Rate i , h k k ( OUT ) ∀ ∈ Ψ ( F ) i , h ( F ) Q Λ Λ ( OUT ) Q ( F ) ( OUT ) i , L i , L ( F ) i , L i i ( OUT ) Service i , L i i ( OUT ) C ( NIC ) µ = Rate i , h i , h

: the set of flows crossing the node Ψ i , h i and leaving it through the NIC h Functi tion Queues Queues NFV node Arrival ( F ) ∑ Λ = λ Rate i , j k ( OUT ) k ∀ ∈ Φ Λ ( OUT ) Q i , j ( F ) i , 1 i , 1 ( F ) Λ Q Service i , 1 i , 1 ( F ) p ⋅ C ( CPU ) C : the transmission rate of the h -th ( NIC ) µ = Rate i , h ( OUT ) ( OUT ) Q i , j i , j i Λ output link of the i -th NFV node ( F ) i , 2 i , 2 ( F ) Λ Q i , 2 i , 2 CPU Outp tput t Queues Queues ( OUT ) ( F ) ( F ) ( OUT ) Q Q Λ Λ i , j i , j i , h Arrival i , h ( OUT ) ∑ Λ = λ Rate i , h k k ( OUT ) ∀ ∈ Ψ ( F ) i , h ( F ) Q Λ Λ ( OUT ) Q ( F ) ( OUT ) i , L i , L ( F ) i , L i i ( OUT ) Service i , L i i ( OUT ) C ( NIC ) µ = Rate i , h i , h

A non-NFV node can be non-NFV node modeled as a set of output queues, one for each output ( OUT ) link Λ ( OUT ) Q i , 1 i , 1 ( OUT ) ( OUT ) Q Λ i , 2 i , 2 Outp tput t Queues Queues ( OUT ) Arrival ( OUT ) Q Λ ( OUT ) ∑ Λ = λ i , h i , h i , h k Rate k ∀ ∈ Ψ i , h Service ( OUT ) ( OUT ) Q Λ ( OUT ) C ( NIC ) µ = ( OUT ) i , L ( OUT ) Rate i , L i i , h i , h i

� The whole netw twork can be modeled as a netw twork of queues � Model definiti tion: an N -dimensional conti tinuous- time Markov chain whose sta ti tate te is defined as follo fo llows: ws: ( ) ( ) … S ( t ) S ( t ), , S ( t ) Σ = 1 N where is equal to: S i ( t ) ( ) (NFV Node) - S ( t ) S ( F ) ( t ), … , S ( F ) ( t ), S ( OUT ) ( t ), … , S ( OUT ) ( t ) = i i , 1 ( F ) i , 1 ( OUT ) i , L i , L i i ( ) (non-NFV Node) … - S ( t ) S ( OUT ) ( t ), , S ( OUT ) ( t ) = i , 1 ( OUT ) i i , L i