Hi @hucklebarryrees ,
Databricks is indeed primarily designed as an analytical platform rather than a transactional system. It’s optimized for data processing, machine learning, and analytics rather than handling high-frequency, parallel transactional requests. For instance, SQL Warehouse clusters in Databricks aren’t ideal for environments with a high volume of parallel requests. Databricks generally recommends a concurrency level of around 10 queries at a time per cluster, so heavy transactional loads could face performance limitations.
However, there are ways to use Databricks effectively for customer-facing applications, particularly when it comes to machine learning models. Databricks offers Model Serving, which lets you deploy ML models as REST APIs. This functionality is well-suited for integrating real-time predictions and inferences into customer-facing applications, enabling scalable ML capabilities while still keeping the main Databricks environment optimized for analytics.
This setup allows you to leverage Databricks' strengths in machine learning and data processing within customer-facing solutions without overloading the platform with transactional demand.
