Description:
- CASQL systems
- Social networking applications
- Web applications with high data query rates
Abstract
USC researchers present the Gumball Technique (GT), a solution to detect and prevent race conditions, ensuring consistency between key-value pairs and tabular data. GT is an application-transparent and deadlock-free technique implemented by the KVS that ignores cache put operations to prevent inconsistencies. GT applies to all CASQL applications, reducing software complexity and costs; it significantly reduces observed inconsistencies; it is compatible with off-the-shelf RDBMSs; it adapts to varying system loads without external adjustments; and it avoids clock drift issues.
Benefit
- Simplifies app development and reduces costs
- Eliminates need for individual race condition detection techniques
- Minimizes the occurrence of data inconsistencies
- Enhances the accuracy and reliability of system
Market Application
Social networking apps rely on data queries as they display user profiles, which are updated infrequently but accessed frequently. To enhance performance, these apps use a cache manager, typically a Key-Value Store (KVS), to store optimized representations of key-value pairs from normalized relational data. Using Cache Augmented SQL RDBMS (CASQL) enhances performance by leveraging faster KVS lookup capabilities. Still, existing techniques face challenges including race conditions and maintaining consistency between the KVS and database.
Publications
Gumball: a race condition prevention technique for cache augmented SQL database management systems, Ghandeharizadeh and Yap, 2012.
Stage of Development