An Approximation Algorithm for Path Computation and Function Placement in SDNs Matthias Rost Technische Universität Berlin July 21, SIROCCO 2016 Joint work with Guy Even and Stefan Schmid Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 1
Introduction
Introduction Networking Opportunities Opportunities: Software-Defined Networking (SDN) and Network Functions Virtualization (NFV) [Kreutz et al., 2015] switches links A C B endpoints Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 3
Introduction Networking Opportunities Opportunities: Software-Defined Networking (SDN) and Network Functions Virtualization (NFV) [Kreutz et al., 2015] SDN Controller A C B Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 3
Introduction Networking Opportunities Opportunities: Software-Defined Networking (SDN) and Network Functions Virtualization (NFV) [Kreutz et al., 2015] SDN Controller A C B Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 3
Introduction Networking Opportunities Opportunities: Software-Defined Networking (SDN) and Network Functions Virtualization (NFV) [Kreutz et al., 2015] SDN Controller A C B Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 3
Introduction Networking Opportunities Opportunities: Software-Defined Networking (SDN) and Network Functions Virtualization (NFV) [Kreutz et al., 2015] SDN Controller Gateway Firewall A C B Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 3
Introduction Networking Opportunities Opportunities: Software-Defined Networking (SDN) and Network Functions Virtualization (NFV) [Kreutz et al., 2015] SDN Controller Gateway Firewall Service Chain: 100$ A A C C B Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 3
Introduction Networking Opportunities Opportunities: Software-Defined Networking (SDN) and Network Functions Virtualization (NFV) [Kreutz et al., 2015] SDN Controller Gateway Firewall x86-Server A C B Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 3
Introduction Networking Opportunities Opportunities: Software-Defined Networking (SDN) and Network Functions Virtualization (NFV) [Kreutz et al., 2015] SDN Controller Gateway Firewall x86-Server A C B Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 3
Introduction Networking Opportunities Opportunities: Software-Defined Networking (SDN) and Network Functions Virtualization (NFV) [Kreutz et al., 2015] SDN Controller Gateway Firewall Service Chain: 100$ x86-Server A A C C B Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 3
Introduction Networking Opportunities Opportunities: Software-Defined Networking (SDN) and Network Functions Virtualization (NFV) [Kreutz et al., 2015] SDN Controller Gateway Firewall Service Chain: 100$ x86-Server A A C C B Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 3
Introduction Networking Opportunities Opportunities: Software-Defined Networking (SDN) and Network Functions Virtualization (NFV) [Kreutz et al., 2015] SDN Controller Gateway Firewall Service Chain: 100$ x86-Server A A C C B Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 3
Introduction Informal Problem Definition Path Computation and Function Placement Problem (PCFP) SDN Controller 300$ / 10 Gbps 10 Gbps 10 Gbps A C 5 Gbps A 150$ / 5 Gbps 10 Gbps B C C A B B 100$ / 10 Gbps Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 4
Introduction Informal Problem Definition Path Computation and Function Placement Problem (PCFP) SDN Controller 300$ / 10 Gbps 10 Gbps ? 10 Gbps A C 5 Gbps ? A 150$ / 5 Gbps 10 Gbps B C C ? A B B 100$ / 10 Gbps Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 4
Introduction Informal Problem Definition Path Computation and Function Placement Problem (PCFP) SDN Controller 300$ / 10 Gbps 10 Gbps 10 Gbps A C 5 Gbps ? A 150$ / 5 Gbps 10 Gbps B C C C ? A B B B 100$ / 10 Gbps Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 4
Introduction Informal Problem Definition Path Computation and Function Placement Problem (PCFP) SDN Controller 300$ / 10 Gbps 10 Gbps 10 Gbps A C 5 Gbps ? A 150$ / 5 Gbps 10 Gbps B C C C ? A B A B B 100$ / 10 Gbps Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 4
Introduction Informal Problem Definition Path Computation and Function Placement Problem (PCFP) SDN Controller 300$ / 10 Gbps 0 Gbps 0 Gbps A C 0 Gbps 5 Gbps 0 G b p ? A s 150$ / 5 Gbps 10 Gbps B C C B ? A A B B 100$ / 10 Gbps Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 4
Introduction Informal Problem Definition Path Computation and Function Placement Problem (PCFP) SDN Controller 300$ / 10 Gbps 0 Gbps 0 Gbps A C 0 Gbps 5 Gbps ? 0 G b p A s 150$ / 5 Gbps 10 Gbps B C C ? B ? A A B B 100$ / 10 Gbps Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 4
Introduction Informal Problem Definition Path Computation and Function Placement Problem (PCFP) SDN Controller 300$ / 10 Gbps 0 Gbps 0 Gbps A C 0 Gbps 5 Gbps 0 G b p A s 150$ / 5 Gbps 10 Gbps B C C B ? A B B 100$ / 10 Gbps Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 4
Introduction Informal Problem Definition Path Computation and Function Placement Problem (PCFP) SDN Controller 300$ / 10 Gbps 0 Gbps 0 Gbps A C 0 Gbps 5 Gbps 0 G b p A s 150$ / 5 Gbps 10 Gbps B C C B ? A B B 100$ / 10 Gbps Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 4
Introduction Informal Problem Definition Path Computation and Function Placement Problem (PCFP) SDN Controller 300$ / 10 Gbps 0 Gbps 0 Gbps A C 0 Gbps 5 Gbps 0 G b p A s 150$ / 5 Gbps 10 Gbps B C C B ? A B B 100$ / 10 Gbps Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 4
Introduction Informal Problem Definition Path Computation and Function Placement Problem (PCFP) SDN Controller 300$ / 10 Gbps 0 Gbps 0 Gbps A C 0 Gbps 0 Gbps 0 G b p A s 150$ / 5 Gbps 10 Gbps B C C 5 Gbps A B B 100$ / 10 Gbps Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 4
Introduction Informal Problem Definition Path Computation and Function Placement Problem (PCFP) SDN Controller 300$ / 10 Gbps 0 Gbps 0 Gbps A C 0 Gbps 0 Gbps 0 G b p A s 150$ / 5 Gbps 10 Gbps Achieved pro fi t: 450$ B C C 5 Gbps A B B 100$ / 10 Gbps Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 4
Introduction Definition of PCFP Formal Definition of PCFP Requests Acyclic graph G i = ( X i , Y i ) Substrate Network mapping restrictions Directed network N = ( V , E ) U i : X i ∪ Y i → 2 V ∪ 2 E capacities c : V ∪ E → R ≥ 0 benefit, demand: b i , d i ∈ R ≥ 0 start, target: s i , t i ∈ X i c ( a, d ) = 10 U i (fw) = { a } U i (gw) = { d } d gw fw a s i t i c c ( a ) = 10 x86 U i ( t i ) = { c } U i ( s i ) = { b } b U i (x86) = { c } Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 5
Introduction Definition of PCFP Formal Definition of PCFP Requests Substrate Network Directed network N = ( V , E ) 150$ / 5 Gbps B C capacities c : V ∪ E → R ≥ 0 c ( a, d ) = 10 U i (fw) = { a } U i (gw) = { d } d gw fw a s i t i c c ( a ) = 10 x86 U i ( t i ) = { c } U i ( s i ) = { b } b U i (x86) = { c } Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 5
Introduction Definition of PCFP Formal Definition of PCFP Requests Acyclic graph G i = ( X i , Y i ) Substrate Network mapping restrictions Directed network N = ( V , E ) U i : X i ∪ Y i → 2 V ∪ 2 E capacities c : V ∪ E → R ≥ 0 benefit, demand: b i , d i ∈ R ≥ 0 start, target: s i , t i ∈ X i Task Find set I ′ ⊆ I of requests to embed and valid realizations ¯ p i for i ∈ I ′ , s.t. p i represents a path from s i � t i ¯ 1 2 capacities of substrate nodes and edges is not violated 3 the profit � i ∈ I ′ b i is maximized. Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 5
Introduction Definition of PCFP Formal Definition of PCFP Request r i Substrate U i (fw) = { a } U i (gw) = { d } d gw fw a s i t i c x86 U i ( t i ) = { c } U i ( s i ) = { b } b U i (x86) = { c } Valid Realizations via Product Networks: pn ( N , r i ) d a gw fw d d c a b a c c b b s i ˆ s i ˆ t i t i d d a a x86 c c b b Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 5
Introduction Definition of PCFP Formal Definition of PCFP Valid Realizations s i - ˆ Any ˆ t i path in pn ( N , r i ) represents a valid realization of request r i . Valid Realizations via Product Networks: pn ( N , r i ) d a d d c a b a c c b b ˆ s i ˆ t i d d a a c c b b Matthias Rost (TU Berlin) Approximating PCFP in SDNs Helsinki, July 2016 5
Recommend
More recommend