Download to read offline
BB-Gen: A Packet Crafter for Data Plane Evaluation
- 1. Fabricio E RodriguezCesen P Gyanesh Kumar Patra Christian Esteve Rothenberg BB-Gen: A Packet Crafter for Data Plane Evaluation Simpósio Brasileiro de Redes de Computadores e Sistemas Distribuídos (SBRC) Campos do Jordão – São Paulo – Brasil Mayo, 2016 Universidade Estadual de Campinas Faculdade de Engenharia Elétrica e de Computação
- 2. Agenda ➔Introduction ➔BB-Gen ➔Examples ➔Limitations and FutureWork ➔Conclusions 2
- 3.
- 4. Introduction 4 Introduction • Flexible fastpath packet processing research in SDN infrastructures • New protocol-independence datapaths • Necessity to evaluate and test network capabilities • Complex network testing tools • Workload generation to simulate the network traffic
- 5. Introduction 5 Packet Generators • Poorsupport of traffic flow characteristics (e.g., packet size and address distribution) • Limited generation of table traces • Limited protocol support
- 6.
- 7. BB-Gen 7 Main Characteristics • Writtenin Python over Scapy library • Open source code • Simple and intuitive CLI interface • Crafts packets for standard and custom protocols. • Simple integration with Traffic Generators (e.g., pktgen-dpdk, NFPA, TCPDUMP, etc.)
- 8. BB-Gen 8 Other Characteristics • Best-caseand worst-case testing • Easy to build, use and extend • Generates table trace files along with every set of PCAPs • Generates more then 1 million of packets • For performance tests (64, 128, 256, 512, 1024, 1280, 1518 Bytes) • Custom packet sizes
- 9. BB-Gen 9 Other Characteristics • Supportsa list of common standard protocols: ➢ Ethernet ➢ IPv4, IPv6 ➢ TCP, UDP ➢ GRE, VXLAN • User defined payload information • Cross-platform tool
- 10.
- 11.
- 12. Examples Generation of 100vxlan traces with packet size of 64B: python main.py -p vxlan -n 100 Generation of random 1k IPv4 traces for performance test: python main.py -p ipv4 -n 1000 -rnip -rnmac -rnport –performance 12
- 13. Examples MACSAD use case: pythonmain.py -u macsad 13
- 14. Examples MACSAD use case: pythonmain.py -u macsad 14
- 15. Examples MACSAD use case: pythonmain.py -u macsad 15
- 16. Examples MACSAD use case: pythonmain.py -u macsad 16
- 17. Examples MACSAD use case: pythonmain.py -u macsad 17
- 18. Examples 18 Using a P414code to autogenerate 1000 traces: python main.py -p4 examples/p4_src/l3_fwd_ipv6.p4 -n 1000 --performance
- 19. Examples 19 Using a P414code to autogenerate 1000 traces: python main.py -p4 examples/p4_src/l3_fwd_ipv6.p4 -n 1000 --performance
- 20. Examples 20 Using a P414code to autogenerate 1000 traces: python main.py -p4 examples/p4_src/l3_fwd_ipv6.p4 -n 1000 --performance
- 21. Examples Using a P414code to autogenerate 1000 traces: python main.py -p4 examples/p4_src/l3_fwd_ipv6.p4 -n 1000 --performance 21
- 22.
- 23.
- 24.
- 25.
- 26.
- 27.
- 28.
- 29. Limitations and FutureWork 29
- 30. Limitations and FutureWork 30 Near future • Addition of more protocols (e.g., ICMP, ICMPv6, ARP, VLAN, etc.) • Packets with a custom sequence of protocols headers • Support for more use-cases (personalized table traces or set of protocols) • Capability to generate a hexdump of the created packet • Integration with P4 runtime ??? Should be here?
- 31.
- 32. Conclusions 32 Conclusions • Simple touse and extend • Candidate for generating trace files for performance and evaluation tests • Simple integration with benchmarking tools • Use-case (MACSAD) demonstrate the facility to create traces covering different complexities and requirements of the project
- 33. Thanks Universidade Estadual deCampinas Faculdade de Engenharia Elétrica e de Computação https://github.com/intrig-unicamp/BB-Gen Simpósio Brasileiro de Redes de Computadores e Sistemas Distribuídos (SBRC) Campos do Jordão – São Paulo – Brasil Mayo, 2016