TL;DR
Databricks offers scalability, data reliability, and seamless integration with Spark analytics, and is ideal for BI reporting. Firebolt is best for serving customer-facing dashboards at scale, due to it's ability to scan small portions of data using smart indexes and compressed storage.
Databricks Photon was built to process large volume of data. However, supporting low latency analytics is hard even with Delta cache. For example, to address lower response times for dashboards, Databricks recommends using Tableau extracts resulting in customers choosing stale data to deliver end user experience.
Firebolt is focused on providing low latency analytics. This enables the ability to leverage Tableau Live Vs Extracts. Delivering the best end user experience with fresh data.
Databricks provides limited concurrency per cluster. To address this shortcoming, customers rely on cluster autoscaling feature to spin up clusters to achieve higher concurrency.
Firebolt engines process queries swiftly. A single engine has the ability to address hundreds of concurrent queries without the need for additional engines. This is primarily due to the efficiency of processing.
Databricks does not leverage indexing to access data. While the Delta Engine leverages partitioning and clustering, no specialized indexes are offered.
Firebolt leverages specialized Indexes that reduce access down to granular data ranges, providing order of magnitude performance improvement enabling sub-second analytics.
Databricks does not support materialized views. While it is possible to materialize data for fast access, this requires the development of additional pipelines and processing. Furthermore, queries will need to be rewritten to point to the materialized data. This adds complexity and processing.
Firebolt’s aggregating indexes are defined once, incrementally auto-synced at ingest and data can be queried at index speed immediately without rewriting queries.
%20logo.png){kind=icon}