{
  "id": 151757,
  "title": "Snowflake to Databricks: what the migration actually costs you",
  "url": "https://urgent.news/2026/08/04/snowflake-to-databricks-what-the-migration-actually-costs-you",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-08-04T23:55:05.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/zephico/snowflake-to-databricks-what-the-migration-actually-costs-you-5e09"
  },
  "original_language": "en",
  "account": "Snowflake and Databricks are two data platforms that offer similar interfaces and SQL capabilities, but they have fundamentally different technologies underneath. One significant difference is in how storage and compute are billed: on Snowflake, storage and compute costs are combined into a single bill, while on Databricks, storage is a separate cost item on the cloud provider's invoice. This distinction can lead to confusion and inaccurate budgeting for organizations migrating data from Snowflake to Databricks.\n\nWhen planning a migration, organizations have three main strategies to consider: lift-and-shift, re-platforming, and re-architecting. The lift-and-shift approach simply replicates the existing Snowflake schema and SQL code in Databricks with minimal changes. This method is the fastest but does not address potential inefficiencies or design choices made for Snowflake's pricing model.\n\nRe-platforming involves converting the schema into a medallion structure, leveraging Delta Lake's capabilities, and redesigning the data layout. This strategy offers better performance and cost efficiency in the long run, as it takes advantage of Databricks' native features like liquid clustering and Photon. However, it is a more time-consuming process and requires a dedicated platform program to justify the investment.\n\nThe re-architecting approach goes further by rebuilding pipelines using Databricks-native patterns, such as streaming ingestion and Delta Live Tables, and leveraging all of Databricks' capabilities. This method provides the best economics and performance advantages but requires the most time and resources to implement, typically only making sense when the organization is planning a long-term platform transformation rather than a one-off migration project.\n\nOne of the biggest challenges in migrating data between these platforms lies in the SQL code itself. While bulk export to Parquet and loading it into object storage can simplify the data transfer process, the actual SQL code may need extensive manual rewriting. This is especially true for JavaScript stored procedures, which have no direct equivalent in Databricks and must be rewritten using PySpark or SQL scripting by individuals with deep knowledge of the original code.\n\nAnother area of potential issues is handling semi-structured data, such as JSON. Snowflake's VARIANT semantics and specific null-handling behavior differ from Databricks, which may result in different query results. Careful validation and testing are required to ensure that semi-structured data is accurately represented during the migration.\n\nTimezone and date arithmetic can also cause discrepancies between Snowflake and Spark, leading to unexpected results in fiscal-calendar or session-window calculations. Thorough testing is necessary to ensure that these calculations remain accurate after the migration.\n\nWhen designing the target platform, it is essential to reconsider the Unity Catalog structure and permissions, as Snowflake's role model does not map directly onto Unity Catalog. Designing the catalog hierarchy, group ownership, and grant boundaries before the migration can save significant time and prevent confusion down the line.\n\nAfter the migration, it is critical to validate the data in three layers: structural checks for errors in data transfer, semantic checks for logic errors in calculations, and financial checks to ensure regulatory and revenue figures are accurately represented. Once validated, the migrated data should be run in parallel with the original Snowflake platform for an extended period, with daily comparisons of outputs to ensure consistency and catch any discrepancies early in the process.\n\nGiven the complexity and potential pitfalls of migrating data from Snowflake to Databricks, organizations should budget for the double bill during the migration period. This budgeting should be explicit and not treated as an afterthought, to avoid budgeting surprises and ensure a smooth transition to the new platform.",
  "summary": "Most Snowflake-to-Databricks migrations get sold on cost and delivered on something else. The credit line item is what gets the project funded, but the teams that finish happy are usually the ones that moved for a different reason: they wanted ML, streaming and GenAI workloads living next to the analytics data instead of shuttling between two platforms. If your only justification is the bill,…",
  "key_points": [
    "Snowflake and Databricks differ in storage/compute billing structure.",
    "Migration strategies: lift-and-shift, re-platforming, re-architecting.",
    "Budget for double billing during migration period."
  ],
  "editors_take": "Organizations migrating from Snowflake to Databricks must consider fundamental differences in technology and billing models to avoid confusion, inaccurate budgeting, and potential pitfalls in performance, cost efficiency, and data representation.",
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}