Every source, how it connects and how it loads.
A full load reads everything on each run. An incremental load reads only the rows changed since the last successful run, tracked by a watermark that doesn't move forward if a load fails, so nothing is skipped on the retry.
| Source | Loads | Connects through |
|---|---|---|
| SQL Server, on-premises | Full or incremental | On-premises data gateway |
| Azure SQL Database | Full or incremental | Fabric connection |
| Fabric Warehouse | Full or incremental | The warehouse's SQL endpoint |
| Dynamics 365 Sales, Customer Service and Power Apps | Full or incremental | Dataverse or Fabric Link |
| Dynamics 365 Finance and Operations | Kept in sync by Microsoft | Fabric Link |
| Dynamics 365 Business Central | Full, append or incremental | Business Central APIs, through the REST API connector |
| Xero | Full or incremental | Xero app, secrets in Key Vault |
| REST APIs returning JSON | Full, append or incremental | API key, bearer token or OAuth2 |
| CSV and Excel files | Full or append | Files landed in your lakehouse |
| SharePoint Online lists | Full or append | Fabric connection |
Incremental loads need the source to record when a row changed, such as a modified date. Where it doesn't, the table is read in full.
Dynamics 365 and Dataverse to Microsoft Fabric.
Each of Microsoft's business applications has its own route into Fabric. The Accelerator uses the one that suits each product, then treats the data the same way as every other source.
Dynamics 365 Sales, Customer Service and Power Apps
Dynamics 365's customer apps and Power Apps keep their data in Dataverse. The Accelerator reads Dataverse tables with Fabric's own Dataverse connector, using a generated query that asks only for the columns it needs and the rows that changed.
- Incremental loads on the modified date every Dataverse table carries
- Authenticates through your Fabric connection
Dynamics 365 Finance and Operations
Finance and Operations data reaches Fabric through Microsoft's Fabric Link, which keeps the selected tables in sync in OneLake. The Accelerator then deals with what makes F&O data hard to report on.
- Records kept separate by legal entity, for multi-company reporting
- Enum codes turned into readable labels
- RecId references resolved into proper relationships
Synced tables use your Dataverse storage, so we keep the table list to what your reporting needs.
Dynamics 365 Business Central
Business Central's APIs are read through the REST API connector, with each API endpoint set up as its own table, so ledgers, customers, items and documents land alongside your other sources.
Native mirroring arrived in Business Central 2026 release wave 2 (version 29.x): a direct sync from the Business Central SQL database into OneLake as Delta tables. Integration with the Fabric Accelerator connector is pending.
SQL Server, Azure SQL and Fabric Warehouse to Microsoft Fabric.
Line-of-business systems, older finance packages and in-house applications often sit on SQL Server. Tables are read with a generated query, so a new table is configuration, and incremental loads pick up only what changed.
SQL Server, on-premises
Read through Microsoft's on-premises data gateway, so the database doesn't need to be opened to the internet. Incremental loads re-read a short window before the last run, to catch rows that were written late.
Azure SQL Database
Read through a Fabric connection that holds the server, database and sign-in, so no connection details are written into pipelines. Changed rows are found from a modified-date column you choose for each table.
Fabric Warehouse
Data already in another Fabric Warehouse is read through its SQL endpoint and brought into the same governed bronze, silver and gold layers as everything else.
Excel, CSV and SharePoint lists to Microsoft Fabric.
Not everything lives in a system. Budgets, targets, mappings and registers often live in spreadsheets and SharePoint lists, and they belong in the same governed layer as the rest of your data.
CSV and Excel files
Files are dropped into a folder in your lakehouse, by hand or by an export from another system, and picked up on the next run. A wildcard reads a month of files at once.
- Excel workbooks and delimited text: comma, pipe, tab or semicolon
- Title rows at the top of hand-made exports can be skipped
- Values that won't convert are recorded and flagged, not silently dropped
SharePoint Online lists
Lists that teams keep in SharePoint, such as trackers, registers and reference data, are read through a Fabric connection and become tables like any other source.
- Each run reads the whole list
- List items only: files in document libraries aren't a source today
REST APIs and Xero to Microsoft Fabric.
Most cloud platforms publish a REST API. The Accelerator's REST connector is configured rather than coded: the endpoint, sign-in, paging and where the records sit in the response are all settings.
Any REST API that returns JSON
One connector covers most SaaS platforms, from CRM and HR systems to ticketing and e-commerce, with each endpoint set up as its own table.
- Sign-in by API key, bearer token or OAuth2, with secrets in your Key Vault
- Paging by offset, page number, cursor or link header
- Incremental loads where the API can filter by a changed date
Xero
Xero's accounting API with the settings already made. One connection covers contacts, invoices, accounts and the other list endpoints, and Xero's rotating sign-in token is renewed on every run.
- Read-only access to your Xero organisation is enough
- Backs off and retries when Xero's limit of 60 calls a minute is reached
- Full or incremental loads
Not on the list? It gets added to the framework, not bolted on.
A new connector follows the same contract as the ones above, so the new source gets the same masking, history, monitoring and audit trail as everything else, and it's reusable from then on. Most requests fall into one of three kinds.
A SaaS platform with an API
If it has a REST API that returns JSON, it can usually be set up on the REST connector with no new code. Where a platform has quirks, we build a version of the REST connector with its settings already made, the way the Xero connector was built.
Another database or data platform
A database such as Oracle, PostgreSQL or MySQL, or another data platform, gets its own connector type in the framework, feeding the same bronze and silver layers. Where Microsoft already replicates a platform into Fabric, as Fabric Link does for Finance and Operations, we can use that route and apply the same controls on top.
Another file type or location
CSV and Excel are covered today. Other formats, such as JSON files, Parquet, XML or fixed-width exports, and files kept somewhere other than your lakehouse, such as Azure storage or an SFTP server, can be added to the file connector rather than handled as one-offs.
- 01
Tell us the source
On a discovery call: what the system is, how it's reached and which tables matter.
- 02
We check the route
API documentation or a sample, how it signs in, and whether it can say what changed.
- 03
Scope and price
The work, the fee and the date are agreed in writing before anything starts.
- 04
Built into the framework
Tested like the shipped connectors, then configured like any other source.
The four-week delivery guarantee covers sources on our connector list. Business Central and new connectors are quoted separately. Read the guarantee →
Microsoft Fabric connectors, answered.
Can Microsoft Fabric connect to Dynamics 365?
Yes, and the route depends on the product. Dynamics 365 Sales, Customer Service and Power Apps are read from Dataverse with Fabric's Dataverse connector. Finance and Operations reaches Fabric through Microsoft's Fabric Link, and the Accelerator adds the legal entity, enum label and relationship handling that F&O reporting needs. Business Central is read through its APIs on the REST API connector. Business Central's native mirroring into OneLake, introduced in its 2026 release wave 2 (version 29.x), is pending integration with the Accelerator's connector.
How do you connect an on-premises SQL Server to Microsoft Fabric?
Through Microsoft's on-premises data gateway, installed on a server in your network that can reach the database. The gateway makes the connection outwards, so the database isn't opened to the internet. The Accelerator then reads tables through a Fabric connection that uses the gateway.
Can Xero data go into Microsoft Fabric?
Yes. The Accelerator has a Xero connector with Xero's settings already made, covering contacts, invoices, accounts and the other list endpoints. It needs read-only access to your Xero organisation, and loads can be full or incremental.
Do we need to build pipelines in Data Factory?
No. Each type of source already has its pipeline, so connecting a new table is a matter of configuration in the management interface: the source, the table and the load type. The Accelerator uses Fabric's own pipelines and connections underneath.
How often can data refresh?
Each group of sources runs on a schedule you choose, such as nightly or hourly, and different sources can run at different times. A source can also belong to more than one schedule, for example a nightly run of everything and an hourly run of your CRM.
Does MPowerUp hold our credentials?
No. Credentials sit in your Fabric connections or your own Azure Key Vault, in your tenant. The Accelerator's configuration records the name of a secret, never the secret itself.
What if our system isn't listed?
We add a connector for it to the framework. A SaaS platform with a REST API can usually be configured without new code; a new database, data platform or file type gets its own connector type. Either way it's scoped and priced before work starts, and it's quoted separately from the four-week delivery guarantee.
Check your sources on a 30-minute call.
Tell us what you run. We'll tell you which connectors cover it, what it would take to add anything that isn't covered, and roughly what it would cost.
A client partner first.