Optimizing LLM Queries in Relational Data Analytics Workloads

TL;DR

This paper introduces algorithms to reorder data in relational analytics workloads to maximize cache reuse, significantly reducing LLM inference costs and improving processing times.

Contribution

It presents novel reordering algorithms that enhance cache efficiency for LLM queries in relational data analytics, a previously underexplored area.

Findings

Up to 3.4x faster job completion times

32% cost savings on LLM inference

Effective for diverse LLM-based queries

Abstract

Batch data analytics is a growing application for Large Language Models (LLMs). LLMs enable users to perform a wide range of natural language tasks, such as classification, entity extraction, and translation, over large datasets. However, LLM inference is highly costly and slow: for example, an NVIDIA L4 GPU running Llama3-8B can only process 6 KB of text per second, taking about a day to handle 15 GB of data; processing a similar amount of data costs around $10K on OpenAI's GPT-4o. In this paper, we propose novel techniques that can significantly reduce the cost of LLM calls for relational data analytics workloads. Our key contribution is developing efficient algorithms for reordering the rows and the fields within each row of an input table to maximize key-value (KV) cache reuse when performing LLM serving. As such, our approach can be easily applied to existing analytics systems and serving platforms. Our evaluation shows that our solution can yield up to 3.4x improvement in job completion time on a benchmark of diverse LLM-based queries using Llama 3 models. Our solution also achieves a 32% cost savings under OpenAI and Anthropic pricing models.