Service Node Replication Chain for Disaggregated Network Offloads
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data centers face challenges in efficiently and cost-effectively providing advanced network services due to the high cost and power consumption of smartNICs, necessitating a disaggregated network architecture that combines purpose-built NICs with service nodes in smart switches.
Innovation Solution
Implementing a network architecture with service nodes configured in a circular replication chain, utilizing a local transport protocol (LTP) for communication between PCIe devices and smart switches, and leveraging service nodes to provide virtualized offloads, reducing the need for full-featured smartNICs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If smartNICs are used to provide advanced network services, then service functionality is improved, but capital expense and operating expense increase significantly
Solution Approach 1:
The patent segments the network service architecture by separating the service processing functionality from the network interface functionality. Service nodes are extracted from individual smartNICs and consolidated into shared service nodes that can be accessed by multiple PCIe devices, thereby reducing the need for every device to have full-featured smartNICs while maintaining service capabilities.
Solution Approach 2:
The patent implements data path state replication where service state information is copied and replicated across multiple service nodes in a circular chain. This replication enables any service node to handle requests for any PCIe device, providing redundancy and load distribution without requiring each device to have dedicated service processing hardware.
2Adaptability or versatility
If smartNICs are used to provide advanced network services, then service functionality is improved, but power consumption increases
Solution Approach 1:
The patent merges the service processing functionality from multiple individual smartNICs into shared service nodes. Instead of each PCIe device having its own power-consuming service processing unit, the service nodes are shared resources that multiple devices can access, thereby consolidating power consumption while maintaining service functionality.
3Quantity of substance
If service nodes are used instead of smartNICs, then cost is reduced, but service state synchronization complexity increases
Solution Approach 1:
The patent implements dynamic service state replication where service state information is continuously synchronized across the circular chain of service nodes. The system dynamically maintains consistency of service state across all nodes through the LTP protocol, allowing any node to serve any PCIe device without manual configuration or complex synchronization logic.
4Reliability
If service nodes are configured in a circular replication chain, then high availability is achieved, but device complexity increases
Solution Approach 1:
The patent creates universal service nodes that can handle requests from any PCIe device in the system. Each service node in the circular chain is configured to provide the same service functionality, making them interchangeable and multi-functional. This universality simplifies the overall system architecture compared to having dedicated service processing in each device, while achieving high availability through the replication chain.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Network traffic flows can be processed by routers, switches, or service nodes. Service nodes may be ASICs that can provide the functionality of a switch or a router. Service nodes can be configured in a circular replication chain, thereby providing benefits such as high reliability. The service nodes can implement methods that include receiving a first packet that includes a source address in a source address field and that includes a destination address in a destination address field. The first packet can be routed to a selected service node that is in the replication chain that includes a plurality of service nodes that are configured for chain replication of a service state information. A service node configured for NAT or some other service can use the first packet to produce a translated packet that can be transmitted toward a destination indicated by the destination address.