Skip to content
TechFabric

Azure Synapse · Two-week readiness sprint

Know what the Synapse estate contains before you pick a date

TechFabric migrates Azure Synapse estates to Databricks with Fabric Airlift. The first two weeks are the Migration Readiness Sprint. Dedicated SQL, serverless SQL, Spark pools and the Azure Data Factory pipelines around them are inventoried separately, because they do not convert as one thing. Lakebridge handles the T-SQL it can. Distribution styles, PolyBase and the ADF graph stay with a person.

A Synapse migration is rarely one platform. Dedicated SQL pools, serverless SQL, Spark pools and the Data Factory pipelines that feed them each fail in a different place. Two weeks on Fabric Airlift inventories all of it and writes down what is in scope.

Source
Azure Synapse
Length
Two weeks
Who it is for
Teams leaving Azure Synapse whose dedicated pool, Spark and ADF still live as three projects
Bench
80 Databricks-certified engineers

What we look at on Synapse

  • Dedicated SQL pools: tables, distributions, stored procedures and the T-SQL that assumes them
  • Serverless SQL databases and the external tables they sit on
  • Spark pools, notebooks and libraries that will not run unchanged on Databricks runtime
  • Azure Data Factory pipelines, triggers and linked services that have to move with the data
  • PolyBase, CETAS and COPY paths, and what they become under Lakeflow or Auto Loader
  • Synapse RBAC plus SQL permissions, and what that map looks like under Unity Catalog

What a person still has to do

  • 01HASH and ROUND_ROBIN distributions have no equivalent to paste. Partitioning on Delta is a design choice.
  • 02Stored procedures in T-SQL stay in the residue. Lakebridge will not turn a procedure into a pipeline.
  • 03ADF graphs become Lakeflow or Databricks Jobs. The dependencies have to be redrawn, not imported.
  • 04Synapse Spark notebooks need a Databricks runtime pass. Libraries and cluster shapes change.
  • 05PolyBase and CETAS become an ingestion path we choose, not a converted statement.

What the two weeks produce

  • 01An inventory where every object is dispositioned as in scope, excluded, or owned by a person
  • 02The exclusions written down and agreed, so the scope still holds when someone questions it in month four
  • 03A dependency map and a wave plan that respects it
  • 04A named owner against every workload surface
  • 05Parity criteria per object type, agreed before any conversion runs
  • 06A hard-object pilot, chosen as the thing most likely to break
  • 07A go or no-go recommendation with the reasoning shown

What makes it faster

The sprint runs on Airlift, our migration accelerator. It composes Databricks Lakebridge for profiling, analysis, SQL conversion and reconciliation, then wraps that in scope acceptance, signed certificates and a reversible cutover. The inventory from these two weeks is the same ledger the migration runs on afterwards.

FAQ

Questions about Synapse to Databricks

Can you migrate Azure Synapse to Databricks?

Yes. Synapse is one of the implemented Airlift routes. The first two weeks are the Migration Readiness Sprint at /databricks/migration-readiness. Dedicated SQL, serverless SQL, Spark and ADF are scoped as separate surfaces, not as one warehouse.

Do we have to move the Data Factory pipelines as well?

Usually, because they are how the data arrives. The sprint names every pipeline that writes into the estate and every one that reads out of it. Some stay on ADF and point at Databricks. Some move. That is an exclusion or a wave, written down, not a surprise in month four.

Our dedicated pool is the only thing we want to leave. Is that enough?

It can be, if the consumers and the ADF graph are willing to follow. The sprint will tell you whether they are. Leaving the Spark pools and moving only the dedicated pool is a legitimate scope, as long as someone signs the exclusions.

Does the sprint commit us to leaving Synapse?

No. It is fixed fee and it ends with a recommendation. The point of week two is to know whether a date is honest.