Using at Scale (1M+ rows in DB)

This document is a guide for using LiteLLM Proxy once you have crossed 1M+ rows in the LiteLLM Spend Logs Database.

Why is UI Usage Tracking disabled?

• Heavy database queries on LiteLLM_Spend_Logs (once it has 1M+ rows) can slow down your API LLM requests. We do not want this happening

Solutions for Usage Tracking

1. Export Logs to Cloud Storage

   • Send logs to S3, GCS, or Azure Blob Storage
   • Log format specification

2. Analyze Data

   • Use tools like Redash, Databricks, Snowflake to analyze exported logs

Need an Integration? Get in Touch

• Request a logging integration on Github Issues
• Get in touch with LiteLLM Founders
• Get a 7-day free trial of LiteLLM here