Industry Challenge and Vision Workshop at the 50th VLDB Conference

Guangzhou, China

August 26, 2024

Overview

Cloud Native databases are designed to fully harness the power of cloud infrastructure, offering dynamic scalability, efficient resource management, and seamless integration with cloud services. These databases are optimized to deliver high performance and resilience in cloud environments by leveraging features like elastic resource allocation and the co-optimization of hardware and software. With the advent of Large Language Models (LLMs), the capabilities of these databases are further enhanced. LLMs enable advanced features like Natural Language to SQL (NL2SQL), allowing users to interact with databases using everyday language, thereby simplifying data access and analysis for users of all technical backgrounds. The convergence of Cloud Native database technology and LLMs is paving the way for more intuitive, powerful, and accessible data services.

To tackle the technical challenges posed by this industry trend, our workshop will adopt a round-table format. Each session will start with a presentation highlighting research problems encountered in industry practice, followed by an open discussion on potential solutions. The workshop is designed as a platform for researchers and practitioners to exchange ideas and share ongoing work without the need for paper submissions or full presentations. Our objective is to collaboratively define the scope of research and guide the community in effectively addressing these emerging challenges.

Agenda

Date: August 26, Monday

Time

Session

Topics

Opening Remarks

-

9:00-10:30

90mins

Foundation Model & Database

From NL2SQL to Data Agent: Opportunities and Challenges

The LLM-based Optimizer for Databases

Foundation Models for Cross-Database Multi-task AI4DB

10:30-11:00

30mins

Coffee Break

-

11:00-12:30

90mins

Data Management for LLM

New Methods for Tabular Data Governance and Analysis

Accurate and Efficient Mining of High Quality Datasets

From Vector Store to Vector DBMS: A RAG View

12:30-14:00

90mins

Lunch Break

-

14:00-15:30

90mins

Cloud Native Database

Heterogeneous Workload Scheduling and Optimization for Serverless Databases

The All-In-One Data System on the Cloud

15:30-16:00

30mins

Coffee Break

-

16:00-17:30

90mins

Hardware & Software Co-Design

Unified SQL Engine for Heterogeneous Computing Unit

Next-generation Data Center Infrastructure Impacts: Public Cloud and Cloud Stack

Closing Remarks

-

Workshop Banquet (2h)

-