Application-Level Connection Handling for Distributed Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional distributed systems are inadequate in detecting and responding to failures quickly, particularly in performance-critical applications like telecommunication data planes and IoT, due to insufficient failure detection granularity, which can lead to delayed recovery and service level agreement violations.
Innovation Solution
A method for application-level connection handling using a connection-less transport protocol, where a connection handling component monitors and reestablishes connections between distributed applications, providing detailed feedback and distinguishing between network and application failures, enabling faster reaction times and improved latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional TCP-based connection handling with orchestration systems is used, then system reliability is maintained, but failure detection time is too slow (seconds to minutes) for performance-critical applications
Solution Approach 1:
The patent segments the failure detection function from the general orchestration system by introducing a dedicated connection handling component at the application level. This component independently monitors connection health using application-specific logic, enabling sub-second detection without waiting for orchestration-level probes. The segmentation allows parallel operation of fast application-level monitoring and slower infrastructure-level monitoring.
Solution Approach 2:
The connection handling component performs preliminary failure detection by continuously monitoring connection health indicators (such as message acknowledgments, round-trip times, and connection state) before the orchestration system's periodic probes would detect the failure. This preliminary action enables the system to react to failures before they propagate to higher-level components.
2Reliability
If TCP protocol with buffering is used, then short network outages are survived, but failure detection at network connection level still takes considerable time
Solution Approach 1:
The patent implements feedback mechanisms where the connection handling component actively monitors acknowledgment messages from the remote end, measures round-trip times, and tracks connection state changes. This continuous feedback loop enables real-time detection of connection degradation and failure without relying on TCP's passive buffering behavior, reducing detection time from seconds to sub-second levels.
3Reliability
If platform-level health monitoring (OpenStack/Kubernetes) is used, then infrastructure failures are detected, but reaction time is on the seconds scale which is insufficient for telecommunication data plane applications
Solution Approach 1:
The patent introduces an intermediary connection handling component that sits between the application and the orchestration system. This intermediary translates application-level connection events into failure detections, enabling the system to bypass the slow orchestration-level monitoring for application-specific failures. The intermediary acts as a mediator that can trigger immediate recovery actions without waiting for the orchestration system's periodic health checks.
4Use of energy by stationary object
If coarse-grained monitoring intervals (5-10 seconds) are used, then system overhead is reduced, but service level agreements cannot be met for performance-critical applications
Solution Approach 1:
The patent implements dynamic monitoring where the connection handling component adjusts its monitoring intensity based on connection state and application requirements. During normal operation, monitoring occurs at standard intervals. When degradation is detected (through increased latency, missed acknowledgments, or error rates), the component intensifies monitoring frequency automatically, enabling adaptive response that meets SLAs without constant high-overhead monitoring.
Data Source
AI summary
A technique for application level based connection handling in a system comprising a plurality of distributed applications (302, 304) is disclosed. A method implementation of the technique is performed by a connection handling component (306) operating at application Level and providing an interface (310) towards a first application (302) of the plurality of distributed applications (302, 304) for establishing a connection with a second application (304) of the distributed plurality of applications (302, 304). The method comprises establishing, upon request of the first application (302) via the interface (310), a connection with the second application (304), and maintaining the connection on behalf of the first application (302) for communication between the first application (302) and the second application (304).