Runtime Variable Sourcing Infrastructure for Decision Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems for sourcing variables for decision engines in online financial platforms are cumbersome, involving significant latency and requiring substantial investment in domain teams and code changes, leading to organizational and processing inefficiencies.
Innovation Solution
A rapid online variable sourcing infrastructure that injects endpoints into domain servers to register and fetch data variables at runtime, allowing for parallel sourcing from multiple domains without duplicating data or incurring significant processing latency, enabling scalable rule infrastructure and computation complexity handling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional variable sourcing methods are used to ensure data availability for decision engines, then data completeness is improved, but data pipeline latency increases and processing efficiency deteriorates
Solution Approach 1:
The system performs preliminary actions by pre-fetching data variables from domain servers and caching them before they are actually needed for decision-making. This allows the decision engine to access previously retrieved data without incurring additional latency, thus maintaining data availability while reducing processing time.
Solution Approach 2:
The patent introduces an intermediary layer (data fetching service and caching mechanism) between the domain servers and the decision engine. This intermediary handles data retrieval and caching operations, allowing the decision engine to access data without direct connections to domain servers, thereby reducing latency while ensuring data completeness.
2Reliability
If multiple data variables are sourced from multiple domain servers to enhance decision-making capability, then decision accuracy is improved, but system complexity increases and processing overhead grows
Solution Approach 1:
The data fetching service is designed as a universal component that can retrieve data from multiple different domain servers using a standardized interface. This multi-functional service handles various data types and sources through a single unified mechanism, reducing system complexity while enabling access to multiple data variables for improved decision accuracy.
Solution Approach 2:
The caching mechanism acts as an intermediary that consolidates data from multiple domain servers into a unified cache structure. This allows the decision engine to access data from multiple sources without managing multiple direct connections, simplifying the system architecture while maintaining the ability to source diverse data variables.
3Speed
If data is cached in memory to reduce latency, then processing speed is improved, but memory resource consumption increases
Solution Approach 1:
The system dynamically adjusts caching parameters such as cache size limits, expiration times, and eviction policies based on available memory resources and access patterns. This allows the system to optimize the balance between processing speed and memory consumption, maintaining fast access to frequently used data while adapting to varying resource availability.
Solution Approach 2:
The caching mechanism implements discard and recovery by evicting less frequently accessed data from memory when resources are constrained, and reloading it when needed. This allows the system to maintain essential data in memory for fast processing while temporarily discarding non-critical data to free up memory resources.
Data Source
AI summary
There are provided systems and methods for a rapid online variable sourcing infrastructure for decision system. Specifically, endpoints may be injected into domain servers serving as a template for sourcing new data variables. When new variables are involved, a first endpoint is injected for registering variables exposed to the domain server, and a second endpoint may be injected to fetch data values at run time, e.g., during the actual decision computation. In this way, data pipeline latency can be avoided during the sourcing process, as the data variables can be sourced from domain servers at run time.