Reasoning about Software Defined Networks Mooly Sagiv Tel Aviv University Thursday (Physics 105) Monday Schrieber 317 Adviser: Michael Shapira Hebrew University

Content Challenges in SDNs Programming Language Abstractions Programming Language Techniques Program Language Tools Other useful tools

Challenges in SDN Programming complexity Reliabilty

The Internet: A Remarkable Story Tremendous success – From research experiment to global infrastructure Brilliance of under-specifying – Network: best-effort packet delivery – Hosts: arbitrary applications Enables innovation in applications – Web, P2P, VoIP, social networks, virtual worlds But, change is easy only at the edge…  4

Inside the ‘Net: A Different Story… Closed equipment – Software bundled with hardware – Vendor-specific interfaces Over specified – Slow protocol standardization Few people can innovate – Equipment vendors write the code – Long delays to introduce new features 5 Impacts performance, security, reliability, cost…

Do We Need Innovation Inside? 6 Many boxes (routers, switches, firewalls, …), with different interfaces.

How Hard are Networks to Manage? Operating a network is expensive – More than half the cost of a network – Yet, operator error causes most outages Buggy software in the equipment – Routers with 20+ million lines of code – Cascading failures, vulnerabilities, etc. The network is “in the way” – Especially a problem in data centers – … and home networks 7

Creating Foundation for Networking A domain, not a discipline – Alphabet soup of protocols – Header formats, bit twiddling – Preoccupation with artifacts From practice, to principles – Intellectual foundation for networking – Identify the key abstractions – … and support them efficiently To build networks worthy of society’s trust 8

Rethinking the “Division of Labor” 9

Traditional Computer Networks 10 Data plane: Packet streaming Forward, filter, buffer, mark, rate-limit, and measure packets

Traditional Computer Networks 11 Track topology changes, compute routes, install forwarding rules Control plane: Distributed algorithms

Traditional Computer Networks 12 Collect measurements and configure the equipment Management plane: Human time scale

Shortest-Path Routing Management: set the link weights Control: compute shortest paths Data: forward packets to next hop

Shortest-Path Routing Management: set the link weights Control: compute shortest paths Data: forward packets to next hop

Inverting the Control Plane Traffic engineering – Change link weights – … to induce the paths – … that alleviate congestion

Avoiding Transient Anomalies Distributed protocol – Temporary disagreement among the nodes – … leaves packets stuck in loops – Even though the change was planned! 16 1 

Death to the Control Plane! Simpler management – No need to “invert” control-plane operations Faster pace of innovation – Less dependence on vendors and standards Easier interoperability – Compatibility only in “wire” protocols Simpler, cheaper equipment – Minimal software 17

Software Defined Networking (SDN) 18 API to the data plane (e.g., OpenFlow) Logically-centralized control Switches Smart, slow Dumb, fast

OpenFlow Networks 19

Data-Plane: Simple Packet Handling Simple packet-handling rules – Pattern: match packet header bits – Actions: drop, forward, modify, send to controller – Priority: disambiguate overlapping patterns – Counters: #bytes and #packets 20 1.src=1.2.*.*, dest=3.4.5.*  drop 2.src = *.*.*.*, dest=3.4.*  forward(2) 3. src= , dest=*.*.*.*  send to controller 1.src=1.2.*.*, dest=3.4.5.*  drop 2.src = *.*.*.*, dest=3.4.*  forward(2) 3. src= , dest=*.*.*.*  send to controller

Controller: Programmability 21 Network OS App #1App #2App #3 Events from switches Topology changes, Traffic statistics, Arriving packets Commands to switches (Un)install rules, Query statistics, Send packets

OpenFlow in the Wild Open Networking Foundation – Creating Software Defined Networking standards – Google, Facebook, Microsoft, Yahoo, Verizon, Deutsche Telekom, and many other companies Commercial OpenFlow switches – HP, NEC, Quanta, Dell, IBM, Juniper, … Network operating systems – NOX, Beacon, Floodlight, Nettle, ONIX, POX, Frenetic Network deployments – Eight campuses, and two research backbone networks – Commercial deployments 22

Dynamic Access Control Inspect first packet of each connection Consult the access control policy Install rules to block or route traffic 23

Seamless Mobility/Migration See host sending traffic at new location Modify rules to reroute the traffic 24

Example Applications Dynamic access control Seamless mobility/migration Server load balancing Using multiple wireless access points Energy-efficient networking Adaptive traffic monitoring Denial-of-Service attack detection Network virtualization 25 See

Challenges of Programming Software Defined Networks 26

Programming OpenFlow Networks OpenFlow makes programming possible – Network-wide view at controller – Direct control over data plane The APIs do not make it easy – Low level of abstraction Challenges – Composition – Concurrency – Correctness – Testing 27 Controller Switches

A Simple example: Firewall A switch connected to two kind of hosts – Trusted hosts via port 1 – Untrusted hosts via port 2 Trusted hosts can freely send packets to untrusted hosts An unstrusted host can only send to a trusted destination which previously sent messages to this host

Firewall 1 2

Firewall Controller Pseudo-code rel trusted(SW, HO) packetIn(s, p, 1)  # packets from trusted hosts send(s, p, 2) # forward the packet to untrusted hosts trusted.insert(s, p.dst) # insert the target of p into trusted controller memory ft.insert(s, p, 1, 2) # insert a per-flow rule to forward future packets packetIn(s, p, 2) -> # packets from untrusted hosts if trusted(s, p.src) then { send(s, p, 1) # forward the packet to trusted hosts ft.insert (s, p, 2, 1) # insert a per-flow rule to forward future packets }

Firewall Controller Pseudo-code(2) packetIn(s, p, 1)  # packets from trusted hosts send(s, p, 2) # forward the packet to untrusted hosts ft.insert(s, src:p.src, 1, 2) # insert a general rule to forward future packets ft.insert(s, dst:p.dst, 2, 1) # allow future packets from 2

A Learning Switch Ttwo hosts (A & B) An OpenFlow switch with 3 ports Host A is connected to port 1 and Host B is connected to port 2 Gradually install forwarding rules Update upon relocation Host A Switch 1 2 Host B

‘A’ sends a message to ‘B’ Host A Switch 1 2 Host B TCP syn dst=B

Forward to the Controller Host A Switch 1 2 Host B TCP syn dst=B 3 send TCP syn dst=B on port 2 send TCP syn dst=B on port 3 learn that A is connected via port 1

‘B’ sends a message to ‘A’ Host A Switch 1 2 Host B TCP ack dst=A

Forward to the Controller Host A Switch 1 2 Host B TCP ack dst=A 3 send TCP ack dst=A on port 1 learn that B is connected via port 2 Install a rule to forward packets from B to A on port 1

‘A’ sends another message to ‘B’ Host A Switch 1 2 Host B dst=B

Forward to the Controller Host A Switch 1 2 Host B dst=B 3 Send dst=B on port 2 Install a rule to forward packets from A to B to port 2

Learning Switch Pseudo-code rel connected (SW, PR, HO) PacketIn(s, p, e) -> connect.insert (s, e, p.src) if connect(s, o, p.dst) then { send (s, p, o) ft.insert(s, p, e, o) } else foreach o in {1, 2, 3} – p # Flood send (s, p, o)

Reasoning about Programs Debugging Testing – Model checking Programming language support – Abstraction – Composition – Ease of use Program verification – Abstraction

Seminar Benefits A cool topic Reasoning Critically read an article Learn to present an article

Seminar Requirements Compilers Read an article (2 weeks) Prepare presentation (1 week) Participate in lectures

Tentative Schedule October 24Michael Shapira Introduction to SDN October 31Mooly SagivIntroduction to Program Reasoning November 7?? November 14??