WAN Optimization Device Lossless Failover via TCP State Mirroring

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

WAN optimization devices face data loss when they acknowledge data locally before it is received by the server, as this data cannot be recovered if the device fails.

Innovation Solution

Implementing a mechanism where WAN optimization devices delay acknowledging data until it is actually received from the target, while maintaining an open TCP window to prevent throttling issues, and using mirrored devices for redundancy to ensure rapid acknowledgement and lossless operation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If WAN optimization devices do local TCP acknowledgement to improve flow, then throughput is improved, but data loss occurs if the device fails

Engineering Contradiction:
ImprovethroughputVSAvoiddata loss
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent applies preliminary action by pre-synchronizing TCP state information between primary and secondary WAN optimization devices before failover occurs. The secondary device maintains up-to-date TCP state through continuous synchronization, so when failover happens, the secondary device can immediately continue data transmission without retransmitting previously acknowledged data, thus preventing data loss while maintaining high throughput

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent uses copying by creating a redundant copy of the TCP state information in the secondary WAN optimization device. This copy is kept synchronized with the primary device through state synchronization mechanisms, allowing the secondary device to take over seamlessly upon primary device failure, ensuring both reliability and continuous high-speed data flow

Inventive Principle:
Principle #26Copying

2Reliability

If WAN optimization devices delay ACK to prevent data loss, then reliability is improved, but throughput decreases due to throttling

Engineering Contradiction:
Improvedata loss preventionVSAvoidthroughput
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent segments the acknowledgment function into two independent parts: local ACK generation for throughput optimization and remote ACK verification for reliability. The local ACK is sent immediately to maintain high throughput, while a separate verification mechanism checks whether the remote end actually received the data, preventing data loss without throttling the overall transmission speed

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary verification mechanism that mediates between the need for immediate ACK (for throughput) and actual data receipt confirmation (for reliability). This intermediary layer allows the system to send ACKs promptly while still verifying data delivery through additional confirmation protocols, resolving the contradiction between speed and reliability

Inventive Principle:
Principle #24Intermediary (Mediator)

3Reliability

If mirrored CNE/LDCM devices are used for redundancy, then reliability is improved, but device complexity increases

Engineering Contradiction:
Improvefailure toleranceVSAvoidsystem configuration
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent merges the failover functionality into the standard TCP state synchronization process between primary and secondary devices. By combining redundancy with the existing state synchronization mechanism, the system achieves failure tolerance without requiring separate complex failover protocols, thus improving reliability while minimizing additional system complexity

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent implements self-service by enabling the secondary WAN optimization device to automatically detect primary device failure and take over data transmission without manual intervention. The automatic failover mechanism monitors the primary device's health and seamlessly transitions to the secondary device when failure is detected, providing high reliability through automated operations rather than complex manual configuration

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP2789136B1Lossless connection failover for single devices
Publication Date: 2018.04.25 BROCADE COMMUNICATIONS SYSTEMS INC
  • EP2789136B1 patent drawingFigure 1
  • EP2789136B1 patent drawingFigure 2
  • EP2789136B1 patent drawingFigure 3

AI summary

The WAN optimization devices delay ACK until the ACK is actually received from the target while opening up the TCP window. When the ACK is received and forwarded, the TCP window is reduced in size. If there are mirrored WAN optimization devices, the original WAN optimization device sends the datagram across the WAN and at the same time sends the datagram to the mirroring WAN optimization device. When the mirroring WAN optimization device ACKs the mirrored datagram, the original WAN optimization device ACKs the host. When the ACK across the WAN is received, the mirroring WAN optimization device gets the ACK forwarded and deletes the mirrored datagram. On a loss of a device, the TCP connection transitions to the mirroring WAN optimization device, which closes the LAN TCP window and transmits all un-ACKed data. Then upon success the mirroring WAN optimization device reopens the LAN TCP window.