Downloaded 126 times
![B4: [Jain et al, SIGCOMM 13] BwE: [Jain et al, SIGCOMM 15]
B4: Google's Software Defined WAN](https://image.slidesharecdn.com/onskeynotevahdat2017-170421142907/75/Networking-Challenges-for-the-Next-Decade-7-2048.jpg)
![B4: [Jain et al, SIGCOMM 13] BwE: [Jain et al, SIGCOMM 15]
B4: From Copy Network to Business Critical
B4traffic
2012 — 2016](https://image.slidesharecdn.com/onskeynotevahdat2017-170421142907/75/Networking-Challenges-for-the-Next-Decade-8-2048.jpg)
Uploaded byOpen Networking Summit
Networking Challenges for the Next Decade
The document discusses the future challenges in networking for cloud infrastructure over the next decade, highlighting the importance of scalability, availability, and manageability in data centers. It emphasizes the transition from cloud 1.0 to cloud 3.0, including advancements such as serverless computing and disaggregated storage. Key strategies include prioritizing network management, automation, and building infrastructure that is resilient and adaptable to future demands.
More Related Content
Similar to Networking Challenges for the Next Decade
Networking Challenges for the Next Decade
- 1. Networking Challenges for theNext Decade Amin Vahdat On behalf of Google Technical Infrastructure and Google Cloud Platform APRIL 4, 2017
- 2. Google Global Cacheedge nodes FASTER (US, JP, TW) 2016 Unity (US, JP) 2010 SJC (JP, HK, SG) 2013 Points of presence >100 Network fiber Google Network More than a collection of data centers
- 3. # # Future regions andnumber of zones Current regions and number of zones 3 3 2 3 3 3 3 3 2 4 3 3 2 Frankfurt Singapore S Carolina N Virginia Belgium London Taiwan Mumbai Sydney Oregon Iowa São Paulo Finland Tokyo Montreal California Netherlands 3 3 33 Google Cloud Regions Adding 11 new regions
- 4. Ubiquitous Cloud...10x Scaling Datacenter Next-gendisaggregation of storage, memory and compute Campus & Metro Cloud regions and campus expansion driving DC interconnect WAN Cloud replication and bandwidth intensive cloud services (e.g., turnkey video, IoT) 10x10x 10x Step Function Disruptions: Bandwidth, Latency, Availability, Predictability
- 5.
- 6. B4 WAN Interconnect Andromeda NFV and network virtualization Jupiter Datacenter Networking ThePillars of SDN @ Google Espresso SDN for public Internet
- 7. B4: [Jain etal, SIGCOMM 13] BwE: [Jain et al, SIGCOMM 15] B4: Google's Software Defined WAN
- 8. B4: [Jain etal, SIGCOMM 13] BwE: [Jain et al, SIGCOMM 15] B4: From Copy Network to Business Critical B4traffic 2012 — 2016
- 9. 10.1.4/24 VNET: 5.4/16 VNET: 192.168.32/24 VNET:10.1.1/24 Load Balancing DoS ACLs VPN NFV Internal Network Andromeda ToR Google Infrastructure Services 10.1.1/24 ToR 10.1.2/24 ToR 10.1.3/24 ToR
- 10. Watchtower Saturn Firehose 1.1 Google Datacenter NetworkInnovation And hardware scale that we could not buy 10 Time Capacity Firehose 1.0 Jupiter 4 Post 1.3Pb/s clusters in 2013
- 11. B4 WAN Interconnect Andromeda NFV and network virtualization Jupiter Datacenter Networking ThePillars of SDN @ Google Public Internet?
- 12. B4 WAN Interconnect Andromeda NFV and network virtualization Jupiter Datacenter Networking ThePillars of SDN @ Google Espresso SDN for public Internet
- 13. Espresso in Context B4 JupiterData Center Google
- 14. Espresso in Context B4 B2 PeeringMetro Jupiter Data Center Google Google
- 15. Espresso in Context B4 Espresso B2 Internet PeeringMetro User Jupiter Data Center Google Google
- 16. Cloud 1.0 Espresso SDN Peering Router Centric Protocols Espresso: Beforeand After Local view Connectivity first Coarse fault recovery Per-metro and global view Application signals Real-time optimization
- 17. Espresso Architecture Overview Label-switched Fabric BGP speaker ExternalPeer Espresso Metro Peering Fabric eBGP Peering
- 18. Espresso Architecture Overview Label-switched Fabric Host Host Host Host Host Host Packet Processor BGP speaker ExternalPeereBGP Peering Espresso Metro Labeled packets specify egress Host Host Host Host Host Peering Fabric
- 19. Espresso Architecture Overview Label-switched Fabric Host Host Host Host Host Host Packet Processor Local Control GlobalController BGP speaker External PeereBGP Peering Espresso Metro Application Signals Labeled packets specify egress Host Host Host Host Host Peering Fabric
- 20. The next wavein computing • Serverless compute in Cloud 3.0 • IoT • Tightly coupled, general purpose distributed computing It’s time to put it all together • Agile Scale • Jitter • Isolation • Performance is great, but only meaningful with availability, manageability, and velocity Next Decade Challenges in Networking
- 21. Virtualization delivers capexsavings to enterprise DCs Cloud 1.0 Last Decade
- 22. Cloud 1.0 Public cloudfrees enterprise from private HW infrastructure Scheduling, load balancing primitives, “big data” query processing Cloud 2.0Cloud 1.0 HW on Demand Now
- 23. Cloud 1.0 Cloud2.0 Serverless compute, real-time intelligence, and machine learning Not data placement, load balancing, OS configuration and patching Cloud 3.0 Compute, not servers The Third Wave of Cloud Computing
- 24. Cloud 2.0 Networking shouldbe aiming for Cloud 3.0 Cloud 3.0Cloud 1.0 The Third Wave of Cloud Computing
- 25. Storage disaggregation: the datacenteris the storage appliance Seamless telemetry and scale up/down Transparent live migration Open Marketplace of services, securely placed and accessed Networking and Cloud 3.0
- 26. Applications+Functions not VMs Policy not middleboxes ActionableIntelligence not data processing SLOs not placement/load balancing/scheduling Networking and Cloud 3.0
- 27. The network willenable next-generation compute infrastructure The network can define next-generation storage infrastructure The right network infrastructure can deliver fundamental new capability Next Decade Challenges in Networking
- 28. How we Prioritize InfrastructureWork Availability Manageability Velocity Stranding Performance
- 29. • First thingsfirst: an insecure infrastructure is an unavailable infrastructure • Stability is more important than efficiency • Network management is critical • Configuration is hard • Automation matters but can be counter to availability “Evolve or Die: High-Availability Design Principles Drawn from Google’s Network Infrastructure.” SIGCOMM 2016. Availability is Paramount
- 30. • Velocity isthe speed of iteration • Retrospective on “Tussle in Cyberspace: Defining Tomorrow’s Internet” • Build for hitless upgrades and self-validation • Debugging and tracing matter ○ Without visibility, performance does not matter • Network fabrics built for expansion and evolution • Launch and Iterate Build for Velocity
- 31. Isolation with reservationsis easy but leads to huge resource stranding ● General-purpose, shared infrastructure to approximate custom-built and reserved Isolation has many components ● Latency, bandwidth, but also the control plane ● Accounting and chargeback are big missing pieces Congestion Control is still really hard ● Rationalizing multiple control loops, flow, endpoint, flow group, Traffic Engineering Isolation is Critical; Stranding is Terrible
- 32. Amdahl’s law appliesand so an incredible, localized optimization that takes any effort to adopt will be ignored 1. Scale 2. Jitter 3. Storage Disaggregation Must optimize from the application all the way to the end user Performance only Matters if End to End
- 33. How we Prioritize InfrastructureWork Availability Manageability Velocity Stranding Performance
- 34. The next waveof computing • Serverless compute in Cloud 3.0 • IoT • Tightly coupled, general purpose distributed computing It’s time to put it all together • Agile Scale • Jitter • Isolation • Performance is great, but only meaningful with availability, manageability, and velocity Next Decade Challenges in Networking
- 35.
- 36. Open Source Google CloudPlatform 36 Google MapReduce Google Bigtable Google Borg Google Borg Google Dremel
- 37. Open Source Google CloudPlatform 37 TCP BBR gRPC Open Config QUIC ...