A Milestone That Stopped Being a Milestone
ADF's connector library crossed 100 in 2022. Microsoft announced it. There were blog posts. There was celebration.
I didn't write a post about it at the time because, honestly, reaching 100 connectors is not interesting news in 2022. And the reason it's not interesting is itself interesting: the connector problem is solved. We don't have connector conversations anymore. The default assumption when I'm designing an ADF pipeline is that the connector exists. If I'm not sure, I check — and it's almost always there.
That's a remarkable shift from 2014, when the connector list was short enough to memorize and the absence of a specific connector was a genuine architecture constraint.
Where the 100+ Connectors Actually Live
Let me make this concrete. The library breaks down across several categories:
Azure Services
All of them. Azure Blob Storage, Azure Data Lake Storage Gen1 and Gen2, Azure SQL Database, Azure SQL Managed Instance, Azure Synapse Analytics (Dedicated and Serverless), Azure Cosmos DB (multiple APIs), Azure Table Storage, Azure Queue Storage, Azure File Storage, Azure Data Explorer (Kusto), Azure Cache for Redis, Azure Search, Azure Event Hubs, and more. If it's an Azure data service, the connector exists.
Databases
SQL Server, Oracle, MySQL, PostgreSQL, MariaDB, IBM DB2, Informix, Teradata, Netezza, Greenplum, Amazon Redshift, Snowflake, Google BigQuery. Every major RDBMS and data warehouse has a native connector. Not "use ODBC and figure it out" — native, with type mapping, bulk loading, and partition support.
SaaS Platforms
Salesforce, Dynamics 365, SAP (multiple variants: SAP ECC, SAP S/4HANA, SAP BW, SAP Table, SAP HANA), ServiceNow, HubSpot, Marketo, Zendesk, Concur, Square, Xero, QuickBooks Online, Shopify, and more. If you're extracting CRM, ERP, or marketing data, there's a connector.
File Formats
CSV, JSON, Parquet, ORC, Avro, Excel, XML, binary. Delta Lake as a first-class format — you can read and write Delta tables directly from Copy Activity, not just from Data Flows. Binary for anything ADF doesn't need to parse.
Cloud Storage
Azure ADLS Gen2, Azure Blob, Amazon S3, Google Cloud Storage, FTP, SFTP, HTTP. Cross-cloud data movement is native.
The REST and ODBC Escape Hatches
The REST connector handles any HTTP API with configurable authentication, pagination, and response parsing. The ODBC connector handles any source with an ODBC driver. These two connectors turn "no native connector" into "moderate configuration effort" rather than "blocked."
The One Pain Point That Remains
The REST connector has improved significantly over the years, but there's still a scenario that sometimes requires dropping to a Custom Activity: REST APIs with multi-step authentication flows.
The pattern: your API requires OAuth 2.0, but the token endpoint is dynamic (it changes per tenant, or per environment), or the token request requires a custom request body format, or the API uses a non-standard OAuth flow. The REST connector's authentication options handle the common cases — client credentials, OAuth 2.0 with standard endpoints, managed identity. When the API deviates from standard patterns, the REST connector's configuration options run out.
My tactical advice before you conclude you need a Custom Activity: check the REST connector's advanced authentication settings in the current version. They've improved each year. What required a Custom Activity in 2019 often doesn't in 2022. Specifically, look at the "Additional headers" and "Token request body" options — these handle more non-standard OAuth flows than the UI makes obvious.
If the REST connector genuinely can't handle it, an Azure Function triggered by a Web Activity is the cleanest solution. Keep the function simple: accept the API parameters, handle the auth, return the data. The ADF pipeline handles orchestration; the function handles the auth complexity.
What the Connector Milestone Actually Means
The connector problem being solved changes how you design ADF architectures. In 2014, the first question in any ADF design conversation was "can ADF connect to X?" In 2022, that question rarely comes up. The conversation starts at "how do we structure the pipeline?" — which is where it should have always started.
That's progress. The connector library being unremarkable is itself the achievement. Table stakes are called table stakes because everyone has them. ADF has them now.
The differentiation story for ADF in 2022 is not the connector library. It's the metadata-driven framework pattern, the CI/CD model, the Data Flows capability, and the integration runtime options. Those are the conversations worth having.
When someone leads with "ADF has 100+ connectors" as a selling point in 2022, ask them what connectors your actual sources need and whether those connectors handle your specific auth and schema requirements. The count is not the story. The capability is.