Visit the post for more.
Overview
The article discusses Facebook's development of the Express Backbone (EBB), a new long-haul network designed to efficiently manage increasing bandwidth demands and separate cross-data center traffic from Internet-facing traffic. It details the design principles, software architecture, and lessons learned during the implementation of this network.
What You'll Learn
1
How to implement a hybrid traffic engineering model using distributed agents and a central controller
2
Why segment routing improves network efficiency and robustness
3
How to utilize sFlow for traffic matrix estimation in large networks
Prerequisites & Requirements
- Understanding of network design principles and traffic engineering
- Familiarity with Open/R and MPLS technologies(optional)
Key Questions Answered
What are the main goals of the Express Backbone network design?
The main goals of the Express Backbone network design include enabling incremental deployment of features, avoiding inefficiencies with distributed traffic engineering, and keeping network state lean by leveraging MPLS segment routing. These goals aim to enhance the network's efficiency and reliability.
How does the hybrid model of traffic engineering work in the EBB?
The hybrid model in the Express Backbone combines centralized control with distributed agents. Local agents manage immediate traffic rerouting during congestion or failures, while the central controller evaluates the overall network state to optimize path allocations, allowing for a balance between quick responses and strategic planning.
What role does the traffic matrix estimator play in the EBB?
The traffic matrix estimator collects sFlow samples from network devices to create a real-time traffic matrix. This matrix is essential for the central controller to compute optimal paths for traffic, ensuring efficient use of network resources and accommodating varying traffic demands.
What are the benefits of using Open/R in the Express Backbone?
Open/R provides a distributed application development platform that facilitates real-time network state discovery and local reaction to topology changes. This enhances the network's responsiveness and allows for efficient routing and traffic management across the Express Backbone.
Technologies & Tools
Software
Open/R
Used for distributed application development and real-time network state discovery in the Express Backbone.
Networking
Mpls
Employed for segment routing to optimize traffic flow and simplify network programming.
Key Actionable Insights
1Implement a hybrid traffic engineering model to enhance network resilience and efficiency.By combining centralized control with distributed agents, you can quickly respond to network issues while maintaining an overall strategic view of traffic management.
2Utilize segment routing to simplify network programming and improve robustness.Segment routing allows for programming label stacks only at the source router, which reduces complexity and enhances the reliability of traffic routing.
3Leverage sFlow for real-time traffic monitoring and matrix estimation.Using sFlow samples can provide insights into traffic patterns, enabling better path allocation and resource management in large-scale networks.
Common Pitfalls
1
Failing to maintain consistent network state across distributed components can lead to inefficiencies and outages.
This issue often arises from the complexity of managing multiple agents and controllers. Implementing robust monitoring and debugging tools can help mitigate these risks.
Related Concepts
Traffic Engineering
Distributed Systems
Network Resilience
Mpls Segment Routing