Database
Logical Design Tuning via Workload-aware E/R-to-Relational Mapping
arXiv cs.DB0 views
arXiv:2611.23632v1 Announce Type: new Abstract: Conceptual data models, like the E/R model, can provide a more stable and intuitive user-facing interface and better logical independence for applications, compared to the relational model which is forced to closely mirror the physical representation. However, this flexibility introduces a challenge as well as an opportunity: a single conceptual schema admits many semantically equivalent relational schemas, whose performance can differ substantially across workloads. Traditional mapping methods use a set of predefined transformation rules or designer-selected mapping strategies, rather than optimizing the relational realization with respect to the expected workload. In this paper, we present a cost-based framework for workload-aware conceptual-to-relational mapping. Given an E/R schema and a representative workload, our framework automatically selects an efficient relational schema by optimizing mapping decisions, including inheritance, entity and relationship representations. User-specified E/R queries are compiled into SQL over the selected relational realization, allowing existing relational query optimizers to determine the physical execution plan. Experiments on real-world-inspired schemas show that mapping choices strongly affect performance; fixed or heuristic strategies exhibit large variability across workloads, whereas our approach consistently achieves competitive performance avoiding pathological cases.
Source:https://arxiv.org/abs/2611.23632PreviousThe Value Semantics Layer: Enabling Machine-Understandable Databases for Agentic AINextCascadeability and the Realization Gap: An Empirical Study of LLM-Powered Semantic Filtering
Related posts
Langfuse vs Phoenix vs LangSmith for Agent Traces
0dev.to
Postgres 19 is out, and it's available on Neon and Databricks Lakebase
0Hacker News
How a 24-minute CI check became a 10-second local check
0dev.to Testing
EDB compra Altinity para integrar PostgreSQL e ClickHouse
0dev.to Kubernetes
Meta Let the Model Write Its Own Memory Prompts. It Beat Humans by 11 Points — Then Distillation Gave Back 13
0dev.to Programming

PostgreSQL 19 New Features Every API Developer Should Know
0dev.to Database