Logical Network State Synchronization Across Datacenters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Centralized management of virtualized environments faces challenges in extending networks across multiple sites, particularly in maintaining synchronized logical network configurations due to limitations in data transfer and management across self-contained but linked datacenters.
Innovation Solution
A method where a global manager identifies and transfers relevant portions of logical network configuration data to a local manager in a series of transactions, including real-time modifications, using asynchronous channels and queues to ensure complete and up-to-date configuration synchronization, even across multiple datacenters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the global manager transfers the full logical network configuration data to the local manager in a single transaction, then the synchronization completeness is improved, but the network disruption time and data transfer overhead increase
Solution Approach 1:
The patent divides the full logical network configuration data into multiple incremental updates. Instead of transferring the entire configuration in one transaction, the global manager sends partial updates sequentially, allowing the local manager to synchronize in smaller steps, thereby reducing the time the local manager is disconnected while ensuring eventual complete synchronization.
Solution Approach 2:
The global manager prepares and queues configuration updates in advance before the local manager comes online. This preliminary action ensures that when the local manager reconnects, it can receive pre-prepared updates without causing extended disruption, as the data is already staged and ready for transfer.
2Productivity
If the global manager transfers only the relevant portion of configuration data, then the data transfer efficiency is improved, but the risk of incomplete synchronization increases
Solution Approach 1:
The local manager sends a synchronization status message to the global manager indicating which configuration data it currently has. The global manager uses this feedback to determine what additional data needs to be transferred, ensuring that only the necessary portions are sent while maintaining complete synchronization. This feedback mechanism prevents both over-transfer and under-transfer of configuration data.
3Loss of time
If the system maintains continuous connection between global and local managers, then the synchronization timeliness is improved, but the network resource consumption and system complexity increase
Solution Approach 1:
Instead of maintaining continuous connections, the system uses periodic reconnection attempts and event-driven updates. The local manager periodically checks for updates or reconnects when configuration changes occur, allowing the system to tolerate disconnections while still achieving timely synchronization when connections are reestablished.
Solution Approach 2:
The patent introduces an asynchronous queue as an intermediary between the global and local managers. This queue buffers configuration updates and manages the transfer process, decoupling the two managers and allowing them to operate independently when disconnected while ensuring updates are eventually delivered when connectivity is restored.
4Adaptability or versatility
If the system handles large volumes of configuration data, then the network coverage and functionality are improved, but the data transfer time and processing overhead increase
Solution Approach 1:
The patent segments large volumes of configuration data into smaller manageable units that can be transferred incrementally. This segmentation allows the system to handle extensive network configurations across multiple datacenters without requiring prohibitively long transfer times, as data is sent in chunks rather than as one large block.
Solution Approach 2:
The global manager pre-processes and stages large configuration datasets before transfer, organizing them into ready-to-send units. This preliminary preparation reduces the actual transfer time when data needs to be synchronized, as the data is already structured and buffered for efficient transmission.
Data Source
AI summary
Some embodiments provide a method for a global manager that manages a logical network configuration for multiple datacenters that each have a local manager for managing the logical network configuration within the datacenter. Based on detecting that a connection to a particular local manager of a particular datacenter has been restored after a period of unavailability, the method identifies a portion of the logical network configuration that is relevant to the particular datacenter. In a series of transactions, the method transfers the identified portion of the logical network configuration to the particular local manager. During the series of transactions, the method identifies modifications to the identified portion of the logical network configuration to be included in the series of transactions. Upon completion of the series of transactions, the method transfers a notification to the particular local manager indicating completion of the series of transactions.