The Redesigned BaBar Event Store: Believe the Hype
Adeyemi Adesanya, Jacek Becla, Daniel Wang
Abstract
As the BaBar experiment progresses, it produces new and unforeseen requirements and increasing demands on capacity and feature base. The current system is being utilized well beyond its original design specifications, and has scaled appropriately, maintaining data consistency and durability. The persistent event storage system has remained largely unchanged since the initial implementation, and thus includes many design features which have become performance bottlenecks. Programming interfaces were designed before sufficient usage information became available. Performance and efficiency were traded off for added flexibility to cope with future demands. With significant experience in managing actual production data under our belt, we are now in a position to recraft the system to better suit current needs. The Event Store redesign is intended to eliminate redundant features while adding new ones, increase overall performance, and contain the physical storage cost of the world's largest database.
Create a lesson
Related papers
Knowledge Graph Lakehouse: A Cloud-Native Architecture for Managing Enterprise Knowledge Graphs at Scale
Bashar Ahmad, Christoph G. Schuetz, Loris Bozzato
ZeroGR: Filtering-First Constrained Decoding for Generative Recommendation
Xiangrui Yang, Jiawei Yang, Liang Zeng et al.
Order-Dependent Complexity of Enumeration and Direct Access for Conjunctive Queries on Dynamic Data
Antoine Amarilli, Nofar Carmeli, Clémence Dumoulin
Text-to-SQL Models Make Choices Without Saying So
Yichen Xu, Shipei Lin, Chenhao Ma
Navigating Runtime Dynamism in Semantic Query Serving
Hengrui Zhang, Xupeng Miao
SQLoom: Set-Contrastive Learning of Syntactic, Semantic, and Logical Dependencies for Text-to-SQL
Hao Fu, Degang Sun, Wan Wei et al.