Analytics Data integration field-manual-complete

Linked service

A linked service is the connection profile that a Data Factory or Synapse pipeline uses to reach another system. It tells the pipeline where the data store or compute service is, how to authenticate, and which integration runtime should connect. In plain English, it is not the data itself. It is the reusable connection setup behind datasets, copy activities, lookups, and transformations. If the linked service is wrong, the pipeline usually cannot connect, read, write, or run securely.

Aliases
No aliases mapped yet
Difficulty
fundamentals
CLI mappings
9
Last verified
2026-05-16

Microsoft Learn

Microsoft Learn describes linked services as connection definitions used by Azure Data Factory and Azure Synapse Analytics. They define how pipelines connect to data stores or compute services, including endpoints, authentication, integration runtime, and connector-specific settings.

Microsoft Learn: Linked services in Azure Data Factory and Azure Synapse Analytics2026-05-16

Technical context

Technically, linked services are JSON-defined objects in Azure Data Factory and Azure Synapse Analytics. They represent connections to storage accounts, SQL databases, REST APIs, SaaS systems, compute environments, Key Vault, and other services. A linked service can use keys, secrets, service principals, managed identities, or Key Vault references, and it may be tied to an Azure, self-hosted, or managed virtual network integration runtime. Pipelines and datasets reference linked services rather than duplicating connection information everywhere.

Why it matters

Linked service matters because it is often the real cause of pipeline failures, security exposure, and environment drift. A pipeline may look correct, but if the linked service points to the wrong server, uses an expired secret, misses a private endpoint, or runs on the wrong integration runtime, the workflow fails or reaches the wrong data. Linked services also shape deployment quality because development, test, and production need different endpoints without rewriting every activity. Good linked service design keeps connections reusable, secure, parameterized where appropriate, and easy to validate. Bad design spreads credentials and endpoint logic across many pipelines. The stronger pattern is to assign ownership and evidence before Linked service becomes a hidden production dependency.

Where you see it

Signals, screens, and Azure surfaces where this term usually becomes operational.

Signal 01

In Data Factory or Synapse authoring, linked services appear under Manage as reusable connections that datasets, activities, triggers, and notebooks reference during incident, audit, and change reviews.

Signal 02

In pipeline failures, they appear through connection errors, expired credentials, wrong endpoints, Key Vault lookup failures, integration runtime problems, or blocked network access during incident, audit, and change reviews.

Signal 03

In CI/CD templates, they appear as JSON objects whose endpoint, secret, database, or runtime properties often differ between development, test, and production during incident, audit, and change reviews.

When this becomes relevant

Specific situations where this term helps solve real Azure design, operations, migration, security, reliability, cost, or governance problems.

  • Connecting pipelines to storage, SQL, REST, or SaaS systems.
  • Centralizing authentication for repeated data movement.
  • Moving endpoint values through CI/CD safely.
  • Troubleshooting failed pipeline connectivity.

Real-world case studies

Different enterprise-style examples that show the term being used to hit measurable objectives.

Case study 01

ADF connection consolidation

Scenario, objectives, solution, measured impact, and takeaway.

Scenario

Atlas Foods had 48 nearly identical Data Factory SQL linked services across development, test, and production factories.

Business/Technical Objectives
  • Reduce duplicated connection definitions by at least half
  • Move secrets out of linked-service JSON
  • Standardize private connectivity through managed virtual network
  • Improve deployment reliability across environments
Solution Using Linked service

The data engineering team redesigned linked services around reusable connection patterns. SQL credentials moved into Key Vault, managed identity handled secret retrieval, and environment-specific server names were handled through deployment parameters. Each linked service referenced the correct integration runtime for private connectivity. Operators exported factory JSON before and after the change to confirm endpoint and authentication consistency. The team also recorded which pipelines depended on the connection, what identity or secret it used, and which network path carried traffic. That dependency map helped release, security, and operations teams troubleshoot connection failures without opening every activity one at a time. A final checkpoint compared expected business outcome, technical health, rollback readiness, monitoring evidence, and owner signoff before the change was accepted into steady-state operations, added to the production runbook, and reviewed with support staff. The team also added an operator checklist that captured owner, scope, approval record, telemetry to watch, rollback criteria, and the evidence required for the first production review. They scheduled a follow-up after the first successful run so platform engineers could compare expected behavior with actual metrics, cost signals, access records, and support tickets instead of treating deployment as the finish line.

Results & Business Impact
  • Linked services dropped from 48 to 14
  • Secret values disappeared from deployment templates
  • Failed environment promotions decreased by 70%
  • Private SQL connectivity passed validation in all factories
Key Takeaway for Glossary Readers

Linked services are connection architecture, so simplifying them improves security, deployment quality, and troubleshooting.

Case study 02

Retail data source migration

Scenario, objectives, solution, measured impact, and takeaway.

Scenario

ShopBeacon moved supplier feeds from on-premises SFTP to Azure Blob Storage and needed pipelines to change sources without rewriting every activity.

Business/Technical Objectives
  • Switch ten pipelines to the new cloud source
  • Keep cutover rollback available for two weeks
  • Avoid exposing supplier credentials in factory JSON
  • Maintain nightly ingestion SLA
Solution Using Linked service

Architects created a new Blob Storage linked service using managed identity and Key Vault references. Existing datasets were adjusted to reference the new service while pipeline structure remained mostly unchanged. During cutover, the old SFTP linked service stayed disabled but available for rollback. Monitoring compared run duration, copy volume, and failure rates across both connection patterns. The team also recorded which pipelines depended on the connection, what identity or secret it used, and which network path carried traffic. That dependency map helped release, security, and operations teams troubleshoot connection failures without opening every activity one at a time. A final checkpoint compared expected business outcome, technical health, rollback readiness, monitoring evidence, and owner signoff before the change was accepted into steady-state operations, added to the production runbook, and reviewed with support staff.

Results & Business Impact
  • All ten pipelines moved with less than 15% activity rewrite
  • Nightly ingestion finished within the existing SLA
  • Supplier passwords were removed from factory configuration
  • Rollback was not needed after two stable weeks
Key Takeaway for Glossary Readers

A linked service lets teams change connection strategy while preserving much of the pipeline design.

Case study 03

Government analytics access review

Scenario, objectives, solution, measured impact, and takeaway.

Scenario

CivicData Office needed to prove which external systems its Synapse pipelines could connect to before a public-sector security review.

Business/Technical Objectives
  • Inventory all pipeline connection targets
  • Verify authentication method for each connection
  • Identify unapproved public endpoints
  • Create review evidence without opening every pipeline manually
Solution Using Linked service

The operations team exported Synapse and Data Factory linked-service definitions, grouped them by connector type, and matched each endpoint to an approved system catalog. Linked services using direct secrets were prioritized for Key Vault migration. Network review confirmed which services used self-hosted integration runtime and which used private endpoints. The evidence package linked each pipeline dependency to a reviewed connection object. The team also recorded which pipelines depended on the connection, what identity or secret it used, and which network path carried traffic. That dependency map helped release, security, and operations teams troubleshoot connection failures without opening every activity one at a time. A final checkpoint compared expected business outcome, technical health, rollback readiness, monitoring evidence, and owner signoff before the change was accepted into steady-state operations, added to the production runbook, and reviewed with support staff.

Results & Business Impact
  • Security review scope was completed in three business days
  • Seven unapproved legacy endpoints were retired
  • Twelve direct-secret services were queued for Key Vault migration
  • Pipeline owners received a clear dependency map
Key Takeaway for Glossary Readers

Linked services give auditors and operators a concrete inventory of what data workflows can reach.

Why use Azure CLI for this?

Azure CLI is useful for listing factories, exporting ARM templates, inspecting linked-service JSON through REST-style commands, and checking related integration runtime or network resources. The portal is fine for one connection, but CLI and templates make environment comparison, evidence collection, and deployment review much faster.

CLI use cases

  • Inventory linked services in a factory or workspace before migration or cleanup.
  • Compare development and production linked-service JSON for endpoint, runtime, and authentication drift.
  • Validate related Key Vault, managed identity, firewall, and integration runtime settings during troubleshooting.
  • Collect connection configuration evidence before credential rotation, private endpoint rollout, or CI/CD deployment.

Before you run CLI

  • Confirm the factory or Synapse workspace, resource group, subscription, and expected environment before exporting configuration.
  • Redact secrets, connection strings, hostnames, and sensitive database names before sharing JSON output.
  • Check whether the linked service uses managed identity, Key Vault, or direct credentials because evidence handling differs.
  • Coordinate with pipeline owners before changing endpoints, integration runtime, or authentication settings.

What output tells you

  • Linked-service JSON shows connector type, endpoint properties, authentication pattern, integration runtime reference, and parameter usage.
  • Related identity output confirms whether the factory or workspace can reach Key Vault and target data services.
  • Integration runtime output explains whether connectivity depends on Azure-hosted, managed virtual network, or self-hosted runtime resources.
  • Output helps identify connection design issues, but pipeline run logs prove how the linked service behaved at runtime.

Mapped Azure CLI commands

Adjacent discovery commands

adjacent
az resource list --resource-group <resource-group> --output table
az resourcediscoverDatabases
az resource show --ids <resource-id>
az resourcediscoverManagement and Governance

Datafactory operations

direct
az datafactory list --resource-group <resource-group>
az datafactorydiscoverAnalytics
az datafactory show --name <factory> --resource-group <resource-group>
az datafactorydiscoverAnalytics
az datafactory create --name <factory> --resource-group <resource-group> --location <region>
az datafactoryprovisionAnalytics
az datafactory pipeline list --factory-name <factory> --resource-group <resource-group>
az datafactory pipelinediscoverAnalytics
az datafactory pipeline-run query-by-factory --factory-name <factory> --resource-group <resource-group> --last-updated-after <utc> --last-updated-before <utc>
az datafactory pipeline-rundiscoverAnalytics
az datafactory trigger list --factory-name <factory> --resource-group <resource-group>
az datafactory triggerdiscoverAnalytics
az datafactory trigger start --factory-name <factory> --resource-group <resource-group> --name <trigger>
az datafactory triggeroperateAnalytics

Architecture context

Technically, linked services are JSON-defined objects in Azure Data Factory and Azure Synapse Analytics. They represent connections to storage accounts, SQL databases, REST APIs, SaaS systems, compute environments, Key Vault, and other services. A linked service can use keys, secrets, service principals, managed identities, or Key Vault references, and it may be tied to an Azure, self-hosted, or managed virtual network integration runtime. Pipelines and datasets reference linked services rather than duplicating connection information everywhere.

Security

Security for linked services centers on credentials, identity, network path, and secret storage. A linked service can contain connection strings, usernames, keys, service principal references, managed identity settings, or Key Vault references. Storing secrets directly increases risk, especially across exported templates and CI/CD pipelines. Strong designs use managed identity or Key Vault where possible, restrict integration runtime network paths, disable unnecessary public access, and monitor credential changes. Operators should also review who can edit linked services because a malicious or careless edit can redirect pipelines to unapproved systems or expose sensitive data through copy activities. Security reviews should record the allowed scope, approval evidence, and exception owner before Linked service expands access.

Cost

Cost impact is indirect but meaningful. Linked services can point pipelines to expensive compute, high-egress paths, premium storage, or inefficient cross-region movement. They can also duplicate connection definitions, making cost ownership harder. A wrong linked service might copy production-scale data from the wrong source, trigger unnecessary transformations, or use a self-hosted runtime that requires maintainable infrastructure. Parameterized and well-owned linked services reduce waste by keeping endpoints intentional. FinOps review should include source and sink region, integration runtime type, data movement volume, connector behavior, and environment mapping. The connection profile is small, but its downstream work can be expensive. Cost reviews should connect Linked service choices to storage, compute, support, or licensing owners.

Reliability

Reliability depends on stable connection design. Linked services should survive credential rotation, firewall updates, private endpoint changes, integration runtime maintenance, and environment promotion. A brittle linked service can break many pipelines at once because multiple datasets and activities depend on it. Reliable designs separate environment-specific values, use clear naming, document dependencies, and test connectivity after deployment. Self-hosted integration runtimes add another reliability layer because host health, network routing, and certificates can affect every linked service using them. Operators should track failures by linked service so repeated pipeline errors are not treated as unrelated incidents. Reliability reviews should prove the normal path, failure path, and rollback path for Linked service.

Performance

Performance depends on the endpoint, connector, integration runtime, network path, and authentication behavior defined by the linked service. A pipeline can be slow because the linked service uses a distant region, overloaded self-hosted runtime, public path with firewall inspection, or connector settings that limit throughput. Performance tuning often starts by checking whether the connection architecture is correct before changing activity settings. Operators should monitor integration runtime utilization, connector errors, data movement throughput, authentication retries, and network latency. A linked service does not process data itself, but it strongly influences how efficiently processing can reach the data. Performance reviews should measure the runtime path affected by Linked service, not only the configuration value.

Operations

Operations teams use linked services during troubleshooting, deployments, credential rotation, and access reviews. They inspect endpoint names, authentication method, Key Vault references, integration runtime, network requirements, and connector settings. CLI, ARM, REST, and exported JSON help compare linked services across environments and detect drift. Day-to-day issues include expired secrets, wrong database names, missing firewall rules, self-hosted runtime outages, and CI/CD substitutions that did not apply. A good runbook starts by testing the linked service, then checks the dataset and activity. That saves time because connection failures often appear as pipeline or data errors. Operations teams should document the owner, normal check, escalation route, and rollback signal for Linked service.

Common mistakes

  • Putting passwords directly in linked services instead of using Key Vault or managed identity where supported.
  • Creating separate linked services for every database when parameters would make one reusable connection safer.
  • Promoting development endpoints into production because CI/CD parameter substitution was incomplete.
  • Troubleshooting only the activity while ignoring expired credentials, blocked network paths, or integration runtime health.