[Paper Review] UDBMS: Road to Unification for Multi-model Data Management

Jiaheng Lu, Zhen Hua Liu+2 more|arXiv (Cornell University)|Dec 23, 2016

Advanced Database Systems and Queries14 references4 citations

TL;DR

This paper proposes UDBMS, a unified database management system designed to natively support multiple data models—such as relational, document, graph, and key-value—within a single backend. By unifying data modeling, query processing, indexing, and transaction management across models, UDBMS enables cross-model queries and consistent transactions, reducing integration complexity and operational overhead.

ABSTRACT

A traditional database systems is organized around a single data model that determines how data can be organized, stored and manipulated. But the vision of this paper is to develop new principles and techniques to manage multiple data models against a single, integrated backend. For example, semi-structured, graph and relational models are examples of data models that may be supported by a new system. Having a single data platform for managing both well-structured data and NoSQL data is beneficial to users; this approach significantly reduces integration, migration, development, maintenance and operational issues. The problem is challenging: the existing database principles mainly work for a single model and the research on multi-model data management is still at an early stage. In this paper, we envision a UDBMS (Unified Database Management System) for multi-model data management in one platform. UDBMS will provide several new features such as unified data model and flexible schema, unified query processing, unified index structure and cross-model transaction guarantees. We discuss our vision as well as present multiple research challenges that we need to address.

Motivation & Objective

• Address the growing challenge of managing diverse data types—structured, semi-structured, and unstructured—across multiple models in modern applications.
• Overcome limitations of polyglot persistence, which requires managing multiple independent databases and incurs high operational and integration costs.
• Eliminate data silos by unifying data models, query processing, indexing, and transaction management into a single, cohesive system.
• Enable cross-model query execution and consistency guarantees to support complex, real-world applications such as customer 360-view, oil & gas analytics, and healthcare data integration.
• Lay the foundation for a scalable, extensible, and flexible multi-model database system that supports evolving data types and schema requirements.

Proposed method

• Propose a unified data model that abstracts across relational, JSON, XML, graph, and key-value models to enable seamless integration.
• Design a unified query processing engine capable of optimizing and executing cross-model queries using a common logical algebra.
• Introduce a unified index structure that supports efficient access across heterogeneous data models, including path, subtree, and neighbor-based indexing.
• Develop cross-model transaction protocols that ensure consistency across different data models, using fine-grained locking mechanisms such as subgraph and subtree locks.
• Enable flexible schema handling by supporting schema-less storage with automatic schema inference for query-time schema richness.
• Design a new benchmark for multi-model databases to evaluate cross-model performance, consistency, and scalability, with support for varied data schemas and query patterns.

Experimental results

Research questions

• RQ1 How can a single database system natively support multiple data models—such as relational, document, graph, and key-value—without sacrificing performance or consistency?
• RQ2 What mechanisms are required to enable efficient cross-model query processing and optimization across heterogeneous data models?
• RQ3 How can cross-model transaction isolation be achieved to ensure consistency when updates span multiple data models?
• RQ4 What are the key challenges in designing a unified index structure that supports diverse data models with low latency and high scalability?
• RQ5 How can a benchmark be designed to rigorously evaluate multi-model database systems across data variety, schema flexibility, and consistency guarantees?

Key findings

• UDBMS enables unified management of diverse data models—relational, JSON, XML, graph, and key-value—within a single backend, significantly reducing integration and operational complexity.
• Cross-model query execution is feasible through a unified logical algebra and query optimizer that can handle heterogeneous data models in a single execution plan.
• Fine-grained locking mechanisms such as subgraph and subtree locks can effectively support cross-model transaction isolation and consistency.
• A global node labeling scheme improves performance in lock management by enabling fast navigation (e.g., getParent, getNextSibling) across different data structures.
• Intra-object sharding is necessary for large graph or tree objects, and distributed sharding strategies for multi-model data require further research to balance load and maintain consistency.
• A new benchmark is being developed to standardize evaluation of multi-model databases, with support for varied schemas, cross-model queries, and consistency metrics.