TL;DR
A OneLake shortcut lets Microsoft Fabric read data that lives somewhere else, without making a copy. The data can sit in another Fabric workspace, in Azure storage, in Amazon S3 or in Google Cloud Storage. The capacity that stores the data pays for storage, and the capacity that reads it pays for the reads. Inside one tenant the reader’s own permissions apply by default, and outside sources use a saved connection instead. Some sources, such as S3 and Google Cloud Storage, are read-only through a shortcut. Use a shortcut when the data can stay where it is, and copy or mirror it when it cannot.
Key Takeaways A shortcut is a pointer, so deleting it leaves the target alone, but deleting files through it can remove them at the source. Fabric supports OneLake, ADLS Gen2, Blob Storage, S3, S3-compatible, GCS, Dataverse, Iceberg and SharePoint targets, plus on-premises sources through a gateway. Same-tenant OneLake shortcuts use passthrough identity by default, while external and cross-tenant shortcuts are always delegated. Consumers pay read transactions from their own capacity, owners pay for storage, and external clouds bill their own egress. Shortcut transformations turn CSV, Parquet, JSON and Excel folders into Delta tables that refresh about every two minutes. Design inside the limits, which include 100,000 shortcuts per item, 10 per OneLake path and five chained shortcuts. Watch on YouTube
Why Most Fabric Deployments Fail at Scale | Amit Chandak, Microsoft MVP
Kanerika’s Chief Analytics Officer and Microsoft MVP, Amit Chandak, walks through why Fabric deployments struggle at scale and what to plan before shared workspaces multiply.
Why a Folder Pointer Changes Your Data Architecture Picture a finance lakehouse holding three years of revenue tables. Marketing needs the same tables for a campaign dashboard by Friday. The usual route is a pipeline, a second copy, a refresh schedule and an argument about which version is right.
A OneLake shortcut replaces that chain with one folder pointer in Marketing’s lakehouse. The tables stay in the finance workspace, the dashboard reads them live, and nobody schedules a load. The convenience comes with rules about identity, billing, source types and limits, and those rules decide whether the design survives production.
What Is a OneLake Shortcut in Microsoft Fabric? A OneLake shortcut is an object that points to another storage location in Microsoft Fabric . Fabric reads that data as if it lived in the shortcut’s own folder.
The place the shortcut points to is the target path. The place where it appears is the shortcut path, as the OneLake shortcuts documentation defines it.
OneLake shortcuts behave like symbolic links. They are independent objects, so deleting a shortcut leaves the target untouched. If someone moves, renames or deletes the target path, the shortcut can break.
Any workload that can read OneLake can read a OneLake shortcut, because it appears as an ordinary folder. Apache Spark, the SQL analytics endpoint, Real-Time Intelligence and Analysis Services all resolve it without special code. Outside Fabric, applications reach shortcuts through the OneLake API, which supports a subset of the ADLS Gen2 and Blob storage APIs.
You can create Microsoft Fabric OneLake shortcuts in lakehouses and KQL databases, either in the Fabric portal or through the REST API. Microsoft told investors in July 2026 that Fabric has over 40,000 paid customers, up more than 60% year over year . Each new workspace in those tenants raises the same design choice between copying data and pointing at it.
Shortcuts are one practical way to get the unified access a data fabric aims for, as Kanerika’s guide to data fabric explains.
How a OneLake Shortcut Behaves Inside a Lakehouse Lakehouses have two top-level folders. The Tables folder holds structured datasets, and the Files folder holds unstructured or semi-structured data. The rules for shortcuts differ between the two, and that difference decides whether a shortcut appears as a table.
Rules for the Tables Folder In Tables, shortcuts can exist only at the top level. OneLake does not support shortcuts in subdirectories of the Tables folder. If the target holds a supported table format such as Delta , the lakehouse syncs the table metadata and recognizes the folder as a table.
Table schemas sync automatically for every shortcut table, new and existing. A schema change at the source therefore reaches the shortcut without a manual update. A shortcut in Tables can link to a single table or to a schema, which is a parent folder holding several tables.
Two details trip teams up. A name with a space never registers as a Delta table, because the Delta format does not support spaces. Schema shortcuts also work only in schema-enabled lakehouses.
Rules for the Files Folder The Files folder has no placement restrictions. You can create shortcuts at any level of the hierarchy, and they can point to internal or external storage holding data in any format.
Table discovery does not happen here. A folder of Parquet files stays a folder until you load it into a table or apply a shortcut transformation.
KQL Databases and Eventhouse A shortcut created in a KQL database appears in the Shortcuts folder and behaves like an external table. You query it with the external_table() function.
Eventhouse database shortcuts add one more option. You can share every subitem of the source database, or pick specific tables, shortcut tables, materialized views and functions.
Automatic schema sync for the database shortcut applies only when you include all subitems. Selecting a subset switches that sync off. Sync for a shortcut table, which tracks column additions, renames and type changes, stays on by default unless you disable it with a KQL command.
KQL shortcuts have a few extra rules in the Real-Time Intelligence documentation . You cannot rename a shortcut created in a KQL database, and you can create only one at a time. A shortcut also cannot sit over a table with an asterisk in a column name.
How Each Engine Reads a Shortcut The same shortcut is readable from every engine, and the syntax is plain. The snippets below follow the Microsoft documentation and use placeholder names.
# Spark: read a shortcut in the Tables folder as Delta
df = spark.read.format("delta").load("Tables/MyShortcut")
display(df)
# Spark SQL: read it as a managed table
df = spark.sql("SELECT * FROM MyLakehouse.MyShortcut LIMIT 1000")
-- T-SQL through the SQL analytics endpoint
SELECT TOP (100) * FROM [MyLakehouse].[dbo].[MyShortcut]
// KQL database: shortcuts are external tables
external_table('MyShortcut') | take 100Outside Fabric, the same data answers at a OneLake URL such as https://onelake.dfs.fabric.microsoft.com/MyWorkspace/MyLakehouse/Tables/MyShortcut/MyFile.csv. Analysis Services reads shortcut tables too. A semantic model in Direct Lake mode reads the shortcut’s Delta files directly, which the Direct Lake section below explains.
Supported OneLake Shortcut Sources and Item Types Shortcuts split into internal ones that point inside OneLake and external ones that reach other storage systems. Both kinds appear as folders, but they differ in authentication, write behavior and limits.
Internal OneLake Shortcuts Internal shortcuts can reference KQL databases, lakehouses, mirrored Azure Databricks catalogs, mirrored databases, semantic models, SQL databases and warehouses. The target can sit in the same item, in another item of the same workspace, or in a different workspace. The item types do not have to match, so a lakehouse can hold a shortcut that points to warehouse data.
A cross-tenant shortcut extends this to a second Fabric tenant. It always uses delegated authentication through an organizational account or service principal in the producer’s tenant. Microsoft points teams to external data sharing instead when the consumer has no identity in the producer’s tenant.
External OneLake Shortcuts As of October 2026, Microsoft Learn lists these external sources. They are Amazon S3, S3-compatible storage, ADLS Gen2, Azure Blob Storage, Dataverse, Google Cloud Storage, Iceberg tables, and OneDrive and SharePoint folders. The Fabric on-premises data gateway adds on-premises and network-restricted locations.
Source How access is authorized What to watch for ADLS Gen2 Organizational account, service principal, workspace identity, SAS or account key Hierarchical namespace must be on. Writes work where the credential allows them. Cross-tenant targets need a service principal or SAS. Azure Blob Storage Organizational account, service principal, workspace identity or SAS Needs a Storage Blob Data Reader, Contributor or Owner role, or a SAS with Read, List and Execute. Amazon S3 IAM user key and secret Read-only. The user needs GetObject, GetBucketLocation and ListBucket on the bucket. S3-compatible storage Key and secret only Microsoft Entra OAuth, service principal and RoleArn are not supported. Google Cloud Storage HMAC key and secret for a service or user account Read-only. Needs storage.objects.get and storage.objects.list, plus storage.buckets.list on the global endpoint. Dataverse Organizational account (OAuth2) or service principal Read-only. The account needs system administrator permission on the Dataverse Managed Lake. OneDrive and SharePoint Organizational account, workspace identity or service principal Folder level only. No subsites, hub sites, personal sites or on-premises SharePoint. Iceberg tables Set by the underlying storage source OneLake generates virtual Delta metadata over Iceberg tables with Parquet data files. On-premises or firewalled storage Standard on-premises data gateway Reaches on-premises S3-compatible storage and S3 or GCS buckets behind a firewall or virtual private cloud.
ADLS Gen2 and Trusted Workspace Access ADLS Gen2 is the most common external target, because many teams already hold a lake there. The storage account needs hierarchical namespace enabled.
The documentation recommends creating the shortcut on a directory at least one level below a container. The copy function does not work on shortcuts that point straight at a container.
Several limits shape the design.
You cannot create more shortcuts inside an ADLS shortcut. Managed private endpoints are not supported. The Copy Blob API does not work through a shortcut. Storage accounts that use Microsoft Purview Data Sharing are excluded. A firewalled storage account needs trusted workspace access . You create a workspace identity and add a resource instance rule on the storage account. Then you create the ADLS or Blob shortcut with the account’s DFS URL.
The feature requires a workspace on a purchased F SKU capacity, so trial capacities do not qualify. It does not work for cross-tenant requests.
Amazon S3, S3-Compatible and Google Cloud Storage S3 and GCS shortcuts are read-only whatever the user’s permissions say. The S3 IAM user needs GetObject, GetBucketLocation and ListBucket on the bucket. Buckets that use S3 Bucket Keys with SSE-KMS also need encrypt and decrypt rights on the key, or reads fail with a 403 Forbidden error.
GCS shortcuts authenticate with an HMAC key and secret for a service account or user account. A bucket-specific endpoint needs storage.objects.get and storage.objects.list, and the global endpoint also needs storage.buckets.list. S3-compatible shortcuts accept key and secret authentication only.
Dataverse, SharePoint and Iceberg Dataverse shortcuts are read-only and need an account with the system administrator permission. SharePoint and OneDrive shortcuts work at folder level, never at file level, and they cannot reach subsites or hub sites. Iceberg shortcuts rely on metadata virtualization , which generates virtual Delta metadata for Apache Iceberg tables whose data files are Parquet.
On-Premises and Network-Restricted Sources The standard on-premises data gateway opens a path to storage the cloud cannot reach directly. Examples include S3-compatible storage hosted on-premises and S3 or GCS buckets placed behind a firewall or virtual private cloud.
Install a current gateway version on a machine that can reach the storage endpoint. Trust the certificate first if the endpoint uses a self-signed one.
Sources that a team plans to move later are good shortcut candidates. The Databricks to Fabric migration guide uses shortcuts for exactly that. The Snowflake to Fabric data sharing breakdown covers the cost side of sharing data across platforms.
Passthrough or Delegated: How OneLake Shortcut Security Works Shortcut security has two layers, the permission on the shortcut path and the permission on the target path. When a user reads through a shortcut, OneLake applies the most restrictive of the two. A user with write access in the lakehouse but read-only access at the target cannot write to the target.
Creating a shortcut needs Write permission on the item where it will live and Read access to the target data. External sources can add requirements inside the source system itself.
Capability Permission on the shortcut path Permission on the target path Create a shortcut Item Write, or OneLake security ReadWrite OneLake security Read (item ReadAll where OneLake security is not yet supported) Delete a shortcut Item Write, or OneLake security ReadWrite Not applicable Read files or folders OneLake security Read OneLake security Read Write to the target Item Write, or OneLake security ReadWrite Item Write, or OneLake security ReadWrite
Passthrough and Delegated Authentication OneLake uses two authentication models, and the shortcut type decides which one applies. Same-tenant OneLake shortcuts use passthrough by default, and you can choose delegated identity when you create them. Cross-tenant OneLake shortcuts and every external shortcut are delegated only.
In passthrough, the shortcut passes the caller’s identity to the target, so each user sees only what that user may see at the source. The source team keeps full control of its permissions.
In delegated mode, an intermediate credential reads the data on behalf of the caller. That credential can be another user’s identity, a service principal or an account key.
Delegated shortcuts let one team manage access for another. A business unit can connect with a fixed identity and then control which of its own users see the data through OneLake security roles. The caller sees the intersection of the security on the consumer side and the security on the producer side.
Delegated shortcuts follow a few rules. Column-level security works on both the producer and consumer side, while row-level security works on the producer side only. A user can belong to only one consumer-side role with column-level security when the producer side also uses row-level security.
Switching an existing shortcut between passthrough and delegated means deleting it and creating it again. Reading external shortcuts through Spark or direct API calls also needs read permission on the item that holds the shortcut path.
Workspace Roles and OneLake Security Roles OneLake security roles do not restrict the workspace Admin, Member and Contributor roles. Those users still need access to both the shortcut path and the target path. They also need Read access to the target to create or update a shortcut.
Viewers, and users with item Read permission, get access through OneLake security roles instead. The guide to OneLake security in Microsoft Fabric covers role design in depth.
The Direct Lake and T-SQL Identity Exception One exception catches security reviewers off guard. Direct Lake over SQL and T-SQL engines in delegated identity mode do not pass the caller’s identity to the shortcut target. They read the data with the item owner’s identity and then apply OneLake security roles to filter what the caller sees.
Permissions you set directly at the target path for the end user are bypassed in that case. The fix is to use Direct Lake over OneLake or T-SQL in user identity mode, as the shortcut security documentation explains.
Who Pays When You Read Through a OneLake Shortcut? OneLake storage is billed pay-as-you-go per GB, and it does not consume Fabric capacity units. Transactions work differently. OneLake adds no separate charge for reads and writes, but those operations consume capacity units from the Fabric capacity that runs the workload.
That corrects a claim many older articles repeat, that OneLake reads and writes are free. Microsoft publishes the rates in its OneLake consumption documentation .
A hot-tier read costs 104 CU seconds per 10,000 operations of up to 4 MB. A write costs 1,626 CU seconds for the same unit.
Where Each Charge Lands With a Shortcut The rule for shortcuts is short. Transactions count against the capacity of the workspace where the shortcut lives, and the capacity that holds the data is billed for storage. A team that reads another team’s lakehouse through a shortcut therefore spends its own capacity units, not the owner’s.
External shortcuts follow a different path. OneLake does not count capacity units for a request that goes to an external source such as ADLS. The external service bills those transactions to you directly.
Microsoft’s own capacity consumption example is easy to reproduce. Workspace2 on Capacity2 holds a shortcut to a Parquet file in Workspace1 on Capacity1. A notebook in Workspace2 reads the file, the read consumes Capacity2 units, and storage stays on Capacity1.
What Happens When a Capacity Is Paused A paused capacity rejects every transaction, though its stored data keeps billing at the per-GB rate. If Capacity2 is paused and Capacity1 is active, the shortcut in Workspace2 cannot read the data, but Workspace1 can. If Capacity1 is paused and Capacity2 is active, Workspace1 stops working while the shortcut in Workspace2 still reads the data.
That is why the consumption documentation says the storage capacity can pause without disrupting consumers in other capacities. A reporting workspace on its own capacity keeps reading the shared tables while a processing capacity sleeps to save cost.
Why Split Workloads Across Capacities Heavy Spark notebooks and SQL pipelines compete with Power BI refreshes when they share one capacity. The practical pattern is one capacity for ingestion and transformation and a second for semantic models and reports. Shortcuts make the split cheap, because the reporting workspace reads the processing workspace’s tables without a second copy.
Sizing each side is a separate exercise. The Fabric capacity planning guide covers SKUs and throttling, and the Fabric pricing explainer covers what each tier costs.
One date matters for planning. Microsoft says that beginning October 15, 2026, it will correct reporting so that OneLake operations from Spark workloads appear fully in OneLake consumption. Rates are not changing, but Spark-heavy tenants may see higher reported consumption.
OneLake Shortcut Caching for Cross-Cloud Reads External reads get expensive when the data sits in another cloud, because the provider charges for egress. Shortcut caching reduces that cost. As OneLake reads files through an external shortcut, it stores them in a cache for the Fabric workspace and answers later reads from the cache.
You set the retention period between 1 and 28 days, and each access resets it. When the remote provider holds a newer version of a file, OneLake serves it from the provider and refreshes the cache. Files larger than 1 GB are not cached.
As of October 2026, caching supports Google Cloud Storage, S3, S3-compatible and on-premises data gateway shortcuts. It also covers on-premises Amazon S3 shortcuts that use Microsoft Entra service principal authentication. ADLS and Blob shortcuts are not on that list, so their reads go straight to storage.
To turn caching on, open Workspace settings, choose the OneLake tab, switch the cache on and pick a retention period. The Reset cache button clears every cached file in that workspace.
OneLake Shortcut Performance and Query Acceleration A OneLake shortcut gives you access to remote data, not local storage speed. Microsoft’s Eventhouse documentation says queries over OneLake shortcuts can be less performant than queries over ingested data. The causes it names are network calls to fetch data from storage and the absence of indexes.
What Drives Shortcut Query Speed Four factors dominate in practice. The first is distance, meaning the region and cloud that hold the target compared with the capacity that reads it. The second is file layout, because many small files cost more requests than a few well-sized Parquet files.
The third is the engine, since a Spark job, a SQL query and a Direct Lake model each read data differently. The fourth is the source itself, which can slow down or throttle requests under load.
Microsoft’s compliance note for query acceleration makes the region point concrete. Data that must stay in a region needs an Eventhouse capacity in that same region.
Query Acceleration for KQL Shortcuts Eventhouse has a purpose-built fix. Query acceleration lets you set a policy on an external Delta table that caches a chosen number of days of data for high-performance queries.
It works over Delta tables from OneLake shortcuts, Amazon S3, Google Cloud Storage and Azure Blob Storage. You switch it on with the Accelerate option when you create the shortcut.
Acceleration has a price and limits. Accelerated data is charged under the OneLake Premium cache meter and adds to the Eventhouse SSD storage you consume. The external table cannot exceed 900 columns, Parquet files above 6 GB compressed are not cached, and performance can suffer beyond 2.5 million data files.
Accelerated shortcuts behave like external tables, so materialized views and update policies are not supported on them.
A Pre-Production Performance Checklist Place the Fabric capacity in the same region as the target when residency or latency matters. Prefer analytics-friendly formats such as Delta or Parquet, and avoid thousands of small files. Test cold and warm reads, because caching changes the second read. Review request limits on the source, since a busy source can throttle analytics users. Estimate cross-cloud egress before production, then switch caching on where the source supports it. Watch OneLake consumption in the Capacity Metrics app, which reports OneLake operations at workspace level, and use OneLake diagnostics for item-level detail. Shortcut Transformations: Turning Files Into Delta Tables A plain shortcut exposes files, and files are not tables. Shortcut transformations close that gap.
They convert CSV, Parquet, JSON or Excel files behind a shortcut into a managed Delta table with Fabric Spark compute. The table then stays in sync with the source folder.
Fabric polls the shortcut every two minutes. New or modified files append or overwrite rows, and deleted files remove their rows. The file transformation documentation adds that subfolders are discovered recursively and that the resulting table works with SQL, Spark and Power BI.
The transformation works with folders from any source that shortcuts support, including ADLS, Blob Storage, Dataverse, S3, GCS, SharePoint and OneDrive. You need the Contributor role or higher, and the folder must hold homogeneous files that share one schema.
Source format Extensions and compression Notes CSV (UTF-8, UTF-16) .csv, .txt, .tsv, .psv, plus .csv.gz and .csv.bz2 .csv.zip and .csv.snappy are not supported. MAP, LIST, STRUCT and raw binary types are not supported. Parquet .parquet with snappy, gzip, lz4, brotli or zstd compression No format-specific limits listed. JSON .json, .jsonl and .ndjson, plus .gz and .bz2 variants Arrays stay arrays. Nested structures flatten up to five levels deep. Excel (preview) .xlsx and .xls A table shortcut merges sheets into one table. A schema shortcut makes one table per sheet. .xlsx is the recommended format.
Limits to Know Before You Rely on Transformations The transformation is read-optimized, so MERGE INTO and DELETE statements on the target table are not supported. It is available only in lakehouses. Subfolder support can be enabled only for new transformations, and once enabled it cannot be turned off.
Excel adds its own limits. Data is always read from cell A1, title banners are ingested as data rows, and leading zeros in identifiers such as ZIP codes are lost. Workbooks with more than 25 sheets are skipped.
Workbooks with Microsoft Purview sensitivity labels cannot be processed. For deeper silver-layer shaping of JSON arrays, Microsoft points to Fabric materialized lake views.
AI-Powered Transformations in Preview A second family applies language processing to .txt files . The built-in transformations summarize text, translate it, label sentiment, redact personal data and extract named entities. The feature is in public preview, accepts .txt input only and is limited to the regions that Azure Language supports.
The practical use is a privacy-compliant table of customer feedback. A PII detection transformation scans the files and redacts names, phone numbers and emails before the data lands in analytics.
OneLake Shortcut, Mirroring, Copy Job or Pipeline? Shortcuts and mirroring are complementary ways to unify data in OneLake without building pipelines. Microsoft’s guidance on shortcuts and mirroring gives a clear dividing line. Shortcuts work only with open formats, while mirroring handles open and proprietary formats, so a proprietary source leaves mirroring as the only option.
Fabric data movement tools come last. Microsoft’s OneLake architecture patterns say to use them when shortcuts and mirroring do not support the source or do not meet the requirement. The table below turns that guidance into a decision aid.
Option Where the data lives Best fit Watch for Shortcut Stays at the source Open-format data in OneLake, ADLS Gen2, S3, GCS or Dataverse that other teams only need to read Source availability, source permissions, and egress billed by an external cloud Shortcut with transformation Files stay at the source and Fabric maintains a Delta table Folders of CSV, Parquet, JSON or Excel that need to be queryable tables Identical file schemas, read-optimized tables, and a two-minute polling cycle Mirroring Replicated into OneLake, or referenced in place for Databricks, Snowflake and Dremio catalogs A whole database or catalog such as Azure SQL Database, SQL Server, Cosmos DB, PostgreSQL, Oracle or SAP Capacity-bound replication, and per-source support limits Copy job or pipeline Physically moved into Fabric Scheduled or incremental loads, heavy transformation, history, and data that must be owned in Fabric You build, schedule and operate the movement
When Not to Use a Shortcut A shortcut is the wrong tool in several situations. The list below reflects design judgment, and each case has a cheaper alternative.
The source is an operational database or a proprietary format, which calls for mirroring. The source cannot tolerate query load from analytics users or has weak availability. You need history or snapshots that the source does not keep. You must write back to an S3, GCS or Dataverse target, which shortcuts treat as read-only. Compliance rules require the data to live in a specific region, so a copy under your control is safer. The infographic below compresses the whole choice into three questions.
Combine the Options Instead of Choosing One Microsoft describes a pattern called mirror once, consume everywhere. A central workspace mirrors an external database, and other workspaces create shortcuts to the mirrored tables instead of mirroring the source again. If mirroring replicates the data into OneLake, shortcuts to it add no extra copy or storage cost.
The same logic applies to ingestion jobs. A Fabric copy job or a Data Factory pipeline lands data that must be owned and shaped. Shortcuts then expose the landed tables to other workspaces without another copy.
Where OneLake File Explorer Fits The OneLake file explorer is a desktop access tool, not an ingestion method. It syncs the OneLake items you can reach into Windows File Explorer, so people can drag local CSV, Excel and Parquet files into OneLake. Shortcuts solve a different problem, which is data placement inside OneLake, and the file explorer guide covers installation and governance.
Direct Lake Semantic Models on OneLake Shortcut Data OneLake shortcuts and Direct Lake fit together. Direct Lake on OneLake supports any Fabric item whose storage resolves to valid Delta tables, natively or through shortcuts. It also never falls back to DirectQuery, as the Direct Lake overview explains.
Direct Lake on SQL differs. It uses the SQL analytics endpoint for table discovery and permission checks, and it reads from a single Fabric data source. It falls back to DirectQuery when it cannot load a Delta table directly, for example when the source is a SQL view.
Shortcut security makes the choice matter, because of the identity exception covered earlier. Direct Lake over OneLake avoids that limitation, which makes it the safer default for models that read shortcut tables. The Direct Lake semantic model build guide walks through the setup.
Create a OneLake Shortcut Step by Step The portal flow is nearly the same for every source. These steps create an internal shortcut between two lakehouses, following the same-tenant shortcut guide .
Open the lakehouse or KQL database and right-click a directory in the Explorer pane. Select New shortcut, New table shortcut or New schema shortcut, depending on your lakehouse settings. Under Internal sources, select Microsoft OneLake and pick the data source. Choose a connection method. Pass-through is the default, and delegated identity uses a configured connection. Expand Files or Tables and select up to 50 subfolders. Review the shortcut names, edit or remove entries, and select Create. The lakehouse refreshes and the shortcut appears in the Explorer pane. Its properties show a Shortcut Type value, which is how you tell a shortcut from a regular folder or table.
Create Shortcuts With the REST API Automation teams create shortcuts through the Fabric REST API, which also has a bulk creation endpoint. The request below creates a OneLake shortcut, and the target object changes for each source type. The shortcutConflictPolicy parameter controls what happens when a shortcut with the same name already exists.
POST https://api.fabric.microsoft.com/v1/workspaces/{workspaceId}/items/{itemId}/shortcuts?shortcutConflictPolicy=CreateOrOverwrite
{
"path": "Files/landingZone",
"name": "MyOneLakeShortcut",
"target": {
"oneLake": {
"workspaceId": "{targetWorkspaceId}",
"itemId": "{targetItemId}",
"path": "Tables/myTablesFolder/someTableSubFolder"
}
}
}The caller needs the OneLake.ReadWrite.All delegated scope. For external targets, the body names a connectionId from a cloud connection you created earlier. It sits inside an object such as adlsGen2, amazonS3 or googleCloudStorage, as the Create Shortcut reference shows.
Build a Two-Workspace Test Environment The cleanest way to learn the behavior is a small test that mirrors a production split. Workspace A holds a lakehouse named Lake01 with five tables called Customer, Date, Geo, Item and Sales. Workspace B starts empty and serves reporting.
Create a new lakehouse in Workspace B, which will never store real data. Choose New shortcut, point it at Lake01 in Workspace A, select the five tables and create the shortcuts. Workspace B can now read the tables, while all the data still lives in Workspace A.
Next, build a semantic model on the shortcut tables in Workspace B. Relate Item ID, Customer ID and Geo ID in Sales to their dimension tables, and relate the sales date to the Date table.
Then build a simple Power BI report, such as total quantity by item category. Saving that report in a third workspace gives you better control over access and versioning.
Test a Paused Source Capacity A paused-capacity test proves the billing behavior in your own tenant. Assign Workspace A to a small paid F SKU capacity such as F2, and keep Workspace B on a different capacity that stays active.
Pause the capacity assigned to Workspace A in the Azure portal. Try to open Lake01 in Workspace A. The open fails, because a paused capacity rejects every transaction. Switch to Workspace B, whose capacity is still running. Open the shortcut lakehouse and the report. The shortcut tables load and the report still shows data. Storage billing for Workspace A continues during the pause, as the consumption documentation confirms. Resume the capacity afterward and confirm that Workspace A opens normally.
Checklist
Microsoft Fabric Checklist
A practical checklist covering data architecture, migration, governance, security and workload optimization, so shortcut designs sit on a sound platform plan.
Get the Checklist → Limits and Mistakes That Break OneLake Shortcuts Most shortcut incidents come from a small set of limits and habits. The table lists the numbers from the current limitations section , so you can design inside them. External targets add one more rule, because ADLS and S3 target paths cannot contain reserved characters from RFC 3986 .
Limit Value or rule Shortcuts per Fabric item Up to 100,000 Unique data source connections per item Up to 100. Internal shortcuts do not count toward this limit. Shortcuts on a single OneLake path Up to 10 Direct shortcuts to shortcut links Up to 5 Subfolders selected in one create flow Up to 50 Characters in names and paths “%” and “+” are not allowed in shortcut names, parent paths or target paths, and non-Latin characters are not supported Table API recognition of a new shortcut Can take up to a minute Cached file size Files larger than 1 GB are not cached Lineage Not available for shortcuts to warehouses and semantic models, and the lineage view covers one workspace only Schema shortcuts Supported only in schema-enabled lakehouses
The Delete Trap Shortcuts do not support cascading deletes. Deleting the shortcut object removes only the shortcut, and the target stays unchanged.
Deleting a file or folder inside a shortcut is different. It deletes that file or folder in the target when you hold write permission there.
Microsoft’s own example uses an ADLS Gen2 shortcut named MyShortcut. Deleting MyLakehouse\Files\MyShortcut removes the pointer. Deleting MyLakehouse\Files\MyShortcut\Foo\Bar removes the Bar directory from the storage account.
The safe habit is a read-only credential unless a team truly needs write-through.
Moved, Renamed or Deleted Targets A shortcut breaks silently when its target moves. Nothing in Fabric stops the source team from renaming a folder. Agree on a change process with each producing team, and treat target paths as published interfaces.
Container-Level Targets and Long Chains Point ADLS shortcuts at a directory below the container, because the copy function fails on a container-level target. Keep chains short as well.
A shortcut can point to another shortcut, up to five direct links. Each hop adds a dependency and another owner to investigate when a read fails.
Region and Workspace Moves A shortcut lives in the same region as the workspace that holds it, so you cannot move a shortcut to another region on its own. Microsoft’s region move procedure recreates the shortcut definitions in a destination workspace. Because shortcuts are references, the process moves the references without moving the underlying data.
Kanerika Service
Microsoft Fabric Consulting and Implementation
Kanerika designs Fabric workspaces, capacities and governance so shared data stays shared without hidden costs or access surprises.
Explore Fabric Services Troubleshooting OneLake Shortcuts Shortcut problems usually fall into five buckets, which are metadata timing, permissions on two paths, identity mode, cache settings and a moved target. The checks below follow that order.
The Shortcut Does Not Appear or Is Not a Table The Table API can take up to a minute to recognize a new shortcut, so refresh before debugging. If it still shows as a folder, check three things.
It must sit at the top level of the Tables folder. Its name must not contain a space. The target must hold a supported table format such as Delta. Access Denied, or Works for One User but Not Another Check both paths first, because OneLake applies the most restrictive permission of the shortcut path and the target path. Then check the identity mode.
A passthrough shortcut tests each user against the target, so two users can see different results. A delegated shortcut tests the connection identity and the OneLake security roles on the shortcut instead.
Source Changes Are Not Visible Schema and metadata updates apply automatically, but propagation time can vary with source performance, cached views and network connectivity. For GCS, S3 and gateway shortcuts, also check the workspace cache settings. The Reset cache button clears every cached file in the workspace.
Direct Lake Cannot Read a Shortcut Table Direct Lake needs valid Delta tables, so confirm that the target folder holds Delta data and that the shortcut sits in Tables. Then check the identity story. Direct Lake over SQL reads with the item owner’s identity, and Direct Lake over OneLake avoids that limitation.
A Shortcut Breaks After a Rename or Move The shortcut stores the target path, so a renamed or moved target leaves it pointing at nothing. Right-click the shortcut, select Manage shortcut, and edit the name, target connection, target location or subpath. Variables from a variable library make that repair a single change per environment.
Talk to Kanerika
Planning a Fabric Shortcut Architecture?
Kanerika maps dataset ownership, capacity billing and access design before the first shortcut goes live, so shared data stays governed and affordable.
Book a Working Session → Running OneLake Shortcuts in Production Domain Ownership and the Three-Workspace Layout OneLake shortcuts are the natural plumbing of a domain-oriented design. Each domain owns its data and publishes certified datasets, and other domains create shortcuts to them instead of copying. Microsoft lists this domain-oriented data mesh as one of its core OneLake patterns .
A three-workspace layout keeps that clean at a smaller scale. Workspace 1 holds raw or curated lakehouses and warehouses, runs ingestion and ETL, and carries the heavy compute. Workspace 2 holds the shortcuts and semantic models, and Workspace 3 holds reports and stays lean.
The layout maps onto bronze, silver and gold layers, so the medallion architecture guide and the Fabric lakehouse guide help with the layer boundaries. The data fabric versus data mesh comparison covers the ownership model, and the workspace design guide covers roles.
Patterns That Work in Practice Cross-workspace sharing. A domain publishes certified tables, and each consumer workspace shortcuts them instead of copying. Cross-cloud analysis. Keep S3 or GCS data in place with a read-only shortcut, and switch caching on to cut repeat egress. Migration staging. Leave data in ADLS Gen2 and shortcut it into a lakehouse while pipelines and notebooks move across. Microsoft’s Synapse Spark migration overview lists OneLake shortcuts in its data access workstream. Document workloads. Point shortcuts at SharePoint or OneDrive folders and apply AI transformations to text files to produce analysis-ready tables. Parameterize Shortcuts Across Environments Test and production should not point at the same data. Workspace variables from a variable library let each environment carry its own shortcut properties, such as the connection ID or the target location.
Microsoft states that variable libraries are the sole supported way to assign variables to shortcuts as of October 2026. REST API assignment is not supported. The deployment pipelines guide shows how those environments get promoted.
Lineage, Catalog and Governance The workspace lineage view draws shortcut relationships between items, but only inside one workspace. Shortcuts to locations outside the selected workspace do not appear.
Lineage for shortcuts to warehouses and semantic models was unavailable in the October 2026 documentation. Teams therefore need a catalog on top, and the OneLake catalog guide explains how it governs shortcuts across workspaces.
Monitoring needs the same attention. A shortcut transformation exposes status and refresh history in the Manage shortcut pane, including row counts and error details. Check that pane before assuming a downstream table is current, and align the checks with your Fabric governance model .
How Kanerika Designs Shortcut-Based Fabric Architectures Kanerika is a Microsoft Solutions Partner for Data and AI with an Analytics specialization. Its Fabric work puts ownership and billing ahead of the shortcut wizard.
The default design keeps ingestion and transformation in one workspace, shortcuts and semantic models in a second, and reports in a third. Each layer runs on a capacity sized for its own workload.
That layered, governed approach shows up in a client delivery. A global heat-transfer manufacturer ran its order journey across three disconnected SAP systems, with no single record of quotes, approvals and deliveries.
Kanerika built an Order Lead Time Analytics platform on Microsoft Fabric. It consolidated roughly 35 source tables into a governed medallion architecture across development, test and production, with 50+ standardized KPIs and four-level row-level security.
The published case study reports 60% less reporting effort across the three SAP systems. It does not hinge on shortcuts, but it shows the workspace separation and security design that shortcut-based setups depend on.
Case Study
60% Less Reporting Effort Across 3 SAP Systems on Fabric
A global heat-transfer manufacturer replaced disconnected SAP reporting with a governed Fabric medallion architecture, 50+ standardized KPIs and four-level row-level security.
Read the Case Study → A second Kanerika engagement tied a business ontology to governed data in OneLake. For a global pharmaceutical enterprise, Kanerika used Fabric IQ Ontology on shared OneLake data . Manual investigation and reconciliation time fell by 70%, and 85+ KPIs were standardized across Power BI semantic models.
Teams moving from Azure Data Factory or Synapse can start from Kanerika’s Azure to Fabric migration service . The migration accelerator, built on FLIP , has run as a native Fabric workload since FabCon 2026. Kanerika reports 80% faster migration timelines, 50% lower migration costs and 65% fewer resources required for it.
Point at Data on Purpose Microsoft Fabric OneLake shortcuts remove data copies. Every responsibility around ownership, cost and access stays in place. The owner still pays for storage, and the consumer still pays for reads.
An external source still bills its own egress. Choose passthrough or delegated identity before the first user connects, and keep chains short.
Use transformations when files must become tables, mirroring when the source is a database, and copy jobs when you must own the data. Teams that follow those rules get the speed of a pointer without surprises on the bill or in the audit.
Frequently Asked Questions
What is the difference between OneLake and data lake? OneLake is the single data lake that every Fabric tenant includes, so there is no storage to provision. A traditional data lake is storage you create, secure and connect to compute yourself. OneLake stores tables in open Delta Parquet or Iceberg formats and answers ADLS Gen2 compatible APIs, so other tools can read it.
What is OneLake in Microsoft Fabric? OneLake is the unified data lake built into every Fabric tenant, and it is the single place for analytics data. It is built on Azure Data Lake Storage, and tables use Delta Parquet or Iceberg formats that any tool can read. A tenant has exactly one OneLake, which you organize with workspaces and domains.
Is Synapse end of life? Support dates differ by Synapse component, so check Microsoft Lifecycle for each one you run. Microsoft documents a migration path from Synapse Spark workloads to Fabric, including how to handle data access with OneLake shortcuts. That guidance lets you plan the move on your own timeline instead of reacting to a deadline.
Is Fabric replacing Azure? No. Fabric is an Azure-based SaaS analytics service, and it runs alongside the rest of Azure. It combines analytics tools that used to be separate Azure services, such as data integration, warehousing and Power BI. Azure still provides virtual machines, networking, databases and application hosting that Fabric does not cover.
How to access Fabric OneLake? You can open OneLake in the Fabric portal and in the OneLake file explorer for Windows. Azure Storage Explorer, Python and PowerShell also reach it through ADLS Gen2 compatible APIs. Access follows workspace roles, item permissions and OneLake security roles. Shortcuts also let other workloads read external data through the same OneLake paths.
What is the difference between data lake and data fabric? A data lake is a storage repository for raw structured and unstructured data. A data fabric is an architecture that connects and governs data across systems using metadata and automation. Microsoft Fabric is a product, and OneLake shortcuts give it fabric-like virtual access, but the data fabric concept is broader than any single product.
Is OneLake a data warehouse? OneLake is the storage layer, and a Fabric warehouse is a separate item that keeps its data in OneLake. The warehouse adds T-SQL, schema enforcement and relational features on top of that storage. A lakehouse is another item type on the same storage, so a single OneLake serves warehouses and lakehouses together.
What is the difference between OneLake and lakehouse? OneLake is the storage foundation for all Fabric data. A lakehouse is a Fabric item that stores tables and files in OneLake and exposes them to Spark and a SQL analytics endpoint. A tenant can hold many lakehouses, and all of them write to the same OneLake, so shortcuts between them never leave the lake.
Is OneLake part of Fabric? Yes. Every Fabric tenant includes OneLake automatically, and you cannot delete it or create a second one. Items such as lakehouses, warehouses and eventhouses store their data there. OneLake storage appears as its own billing line, and reads and writes consume capacity units. There is no separate OneLake product to deploy.
What are the benefits of OneLake? OneLake gives the whole organization one governed lake, so teams stop building separate lakes. Engines share one copy of data without duplication, and OneLake security roles apply across SQL, Spark and Power BI. Shortcuts extend the same view to ADLS, S3 and other sources, and open Delta and Iceberg formats keep the data portable.
What is the difference between Microsoft Fabric and Azure? Azure is Microsoft’s cloud platform with hundreds of separate services. Fabric is a SaaS analytics product that runs on Azure and bundles data integration, engineering, warehousing, real-time analytics and Power BI around OneLake. Fabric uses capacity-based licensing, while Azure services usually bill each service on its own meter, which changes how you budget.
Is Microsoft Fabric the same as Snowflake? Fabric and Snowflake are separate products. Fabric is an end-to-end analytics platform built around OneLake and Power BI. Snowflake is a cloud data platform centered on warehousing, data sharing and its own engines. They can work together, because Fabric can mirror Snowflake databases and read Iceberg tables. Many enterprises compare the two and then connect them.
Is Microsoft Fabric similar to Databricks? Both follow the lakehouse idea and use Delta tables, so the overlap is real. Fabric is a SaaS platform that bundles integration, engineering, warehousing, real-time analytics and Power BI around OneLake. Databricks centers on Spark-based engineering, machine learning and cross-cloud reach. Many enterprises run both, and Fabric can reference mirrored Azure Databricks catalogs.
What is the difference between data lake and lakehouse? A data lake stores raw files with little structure or enforcement. A lakehouse adds table formats such as Delta, which bring ACID transactions, schema enforcement and SQL access to the same lake storage. In Fabric, a lakehouse keeps its tables and files in OneLake, so it behaves like both a lake and a warehouse.
Does Microsoft Fabric require OneLake? Yes. OneLake is built into every Fabric tenant, and Fabric items store their data there by default. You cannot turn it off or swap it for another store. Shortcuts can still reference outside storage, but the data that Fabric items create is stored in OneLake. Plan capacity and governance around that fact.
Does a OneLake shortcut copy data? No. A shortcut is a pointer, so Fabric reads the data from the target and nothing is duplicated at creation. A shortcut transformation is the exception, because it maintains a managed Delta table converted from the source files. Shortcut caching also keeps recently read external files, but only to speed repeat reads and cut egress.
Which data sources support OneLake shortcuts? Shortcuts support other OneLake items, ADLS Gen2, Azure Blob Storage, Amazon S3, S3-compatible storage, Google Cloud Storage, Dataverse, Iceberg tables, and OneDrive and SharePoint folders. The on-premises data gateway adds on-premises and firewalled storage. Internal targets include lakehouses, warehouses, KQL databases, SQL databases, semantic models, mirrored databases and mirrored Azure Databricks catalogs.
Are OneLake shortcuts read-only? It depends on the source. S3, Google Cloud Storage and Dataverse shortcuts are read-only whatever the user’s permissions say. ADLS Gen2 and internal OneLake shortcuts can accept writes when the permissions on both the shortcut path and the target path allow it. Write carefully, because a delete through a shortcut can remove the file at the source.
What permissions are required to create a OneLake shortcut? You need Write permission on the item that will hold the shortcut, or OneLake security ReadWrite, plus Read access to the target data. External sources can add requirements in their own systems, such as storage roles or IAM permissions. The workspace Admin, Member and Contributor roles all satisfy the Write requirement.
What is delegated authentication for OneLake shortcuts? In delegated authentication, a configured connection identity reads the target data instead of the signed-in user. Depending on the source, that identity is an organizational account, service principal, workspace identity, SAS token or key. External and cross-tenant shortcuts always use it, and users see the intersection of their own OneLake security and the connection’s access.
Do OneLake shortcuts work with Power BI Direct Lake? Yes. Direct Lake on OneLake supports any Fabric item whose OneLake storage resolves to valid Delta tables, natively or through shortcuts. One caution applies to identity, because Direct Lake over SQL reads with the item owner’s identity. Direct Lake over OneLake avoids that limitation and is the safer choice for shortcut tables.
Do OneLake shortcuts consume Fabric capacity? Yes. Reads and writes through a shortcut consume capacity units from the capacity of the workspace where the shortcut lives. Storage is billed to the capacity that holds the data. Requests to external sources such as ADLS or S3 do not consume Fabric capacity units, because the external provider bills them directly.
Do OneLake shortcuts cause AWS or Google egress charges? They can. A shortcut to S3 or Google Cloud Storage reads data across clouds, so the provider can bill requests and data transfer to your cloud account. Shortcut caching reduces repeat reads and egress for S3, S3-compatible, Google Cloud Storage and gateway shortcuts. Estimate transfer costs before a large workload reads external buckets daily.
What is OneLake shortcut? A OneLake shortcut is an object in a lakehouse or KQL database that points to data stored elsewhere. Fabric reads the data in place, so nothing is copied. Targets include other OneLake items, ADLS Gen2, Blob Storage, S3, Google Cloud Storage, Dataverse and SharePoint. The shortcut appears as a folder, and deleting it leaves the source data untouched.
What is OneLake shortcut caching? Caching stores files that OneLake reads through an external shortcut in a cache for the workspace. Later reads come from the cache, and a newer version at the source replaces the cached copy. Retention runs from 1 to 28 days. Caching supports S3, S3-compatible, Google Cloud Storage and gateway shortcuts, and it skips files over 1 GB.
What happens if you delete a OneLake shortcut? Deleting the shortcut removes only the pointer, and the target data stays unchanged. Shortcuts do not support cascading deletes. Deleting a file or folder inside the shortcut is different, because it deletes that item in the target when you hold write permission there. If the target itself is deleted or moved, the shortcut breaks.
Can OneLake shortcuts point to other shortcuts? A shortcut can point to another shortcut, up to five direct links. Each hop adds a dependency and another owner to check when a read fails, so shorter chains are easier to run and to audit. ADLS and S3 shortcuts cannot hold further shortcuts inside them, which caps nesting at the source.
How many OneLake shortcuts can you create? Each Fabric item supports up to 100,000 shortcuts and up to 100 unique data source connections. A single OneLake path supports up to 10 shortcuts. Shortcut names and paths cannot contain the percent or plus characters, and non-Latin characters are not supported. Internal shortcuts do not count toward the connection limit.
Can you transform data through a OneLake shortcut? Yes. Shortcut transformations convert CSV, Parquet, JSON and Excel files into managed Delta tables using Fabric Spark, and they sync about every two minutes. AI-powered transformations in preview summarize, translate, label sentiment, redact personal data and extract entities from text files. Both options work in lakehouses only. The target tables are read-optimized, so MERGE and DELETE statements are unsupported.
What is the difference between OneLake shortcuts and mirroring? A shortcut references selected open-format data in place, while mirroring brings a whole database or catalog into Fabric. Mirroring either replicates data into OneLake as Delta tables or, for catalogs such as Databricks and Snowflake, references it in place. A proprietary source format leaves mirroring as the only option, because shortcuts need open formats such as Delta, Parquet or Iceberg.
What is the difference between a OneLake shortcut and a Fabric pipeline? A shortcut exposes data where it already lives. A pipeline or copy job physically moves and reshapes data on a schedule. Choose a shortcut for read access to open-format data. Choose a pipeline when you need history, heavy transformation, regional control or a copy that you own and operate yourself.
What is the difference between OneLake shortcuts and OneLake file explorer? A shortcut is an object inside a lakehouse that points to other storage. The OneLake file explorer is a Windows app that syncs OneLake items into File Explorer so people can browse files and drag local files in. The explorer is a desktop access tool, not an ingestion pattern for shared data.
Can OneLake shortcuts work across Fabric tenants? Yes. A cross-tenant OneLake shortcut reads data from another Fabric tenant through a delegated identity. That identity is an organizational account or a service principal in the producer’s tenant. The producer grants the Viewer workspace role and a OneLake security role with Read access. External data sharing suits partners who have no identity in the producer’s tenant.
Can OneLake shortcuts connect to SharePoint or OneDrive? Yes. OneDrive and SharePoint shortcuts work at folder level only, and they authenticate with an organizational account, workspace identity or service principal. They cannot reach SharePoint subsites, hub sites, personal sites or on-premises SharePoint. AI-powered transformations can then process text files stored in those folders, but nested folders do not work with SharePoint shortcuts.
How is data stored in OneLake? OneLake organizes data by tenant, workspace and item, and every Fabric item writes its data there. Tables are stored as Delta Parquet or Iceberg, while files can be any format. Because OneLake is built on Azure Data Lake Storage, tools reach it through ADLS Gen2 compatible APIs and the OneLake file explorer.
Can you use OneLake shortcuts with KQL databases? Yes. A shortcut in a KQL database appears under Shortcuts and is queried as an external table with the external_table function. You cannot rename KQL shortcuts, and you create them one at a time. Query acceleration can cache shortcut data in the Eventhouse for faster queries, at the cost of OneLake Premium cache storage.
How do you improve OneLake shortcut performance? Keep the target in the same region as the Fabric capacity, store data in Delta or Parquet, and avoid thousands of small files. Turn on shortcut caching for S3, Google Cloud Storage and gateway sources. In an Eventhouse, enable query acceleration for frequently queried shortcut tables. Test cold and warm reads before going live.
Does a OneLake shortcut work when the source capacity is paused? Yes, for internal shortcuts, as long as the capacity hosting the shortcut is active. The paused capacity rejects direct access to its own data, and its storage billing continues. If the consumer capacity is paused instead, reads through its shortcuts stop. Plan separate capacities for processing and reporting to use this behavior.
How do you create a OneLake shortcut in a lakehouse? Right-click a directory in the lakehouse Explorer, select New shortcut, New table shortcut or New schema shortcut, and pick a source. Choose a connection method, select up to 50 subfolders, review the names and select Create. You can also create shortcuts through the Fabric REST API, which includes a bulk creation endpoint.
What is the difference between Direct Lake on OneLake and Direct Lake on SQL? Direct Lake on OneLake reads Delta tables from one or more Fabric sources and never falls back to DirectQuery. Direct Lake on SQL reads through the SQL analytics endpoint of a single source. It falls back to DirectQuery when it cannot load a Delta table directly, such as a SQL view. The choice also affects how shortcut identity works.
What is the difference between Microsoft Fabric and Service Fabric? Microsoft Fabric is a SaaS analytics platform built around OneLake, with data integration, engineering, warehousing and Power BI. Azure Service Fabric is a separate platform for building and running microservices and containerized applications. The names overlap, but the products serve different teams, so a data team and an application platform team will rarely evaluate them together.
Does Microsoft have a data lake? Yes. Microsoft offers OneLake inside Microsoft Fabric and Azure Data Lake Storage Gen2 as standalone Azure storage. OneLake is a managed lake that every Fabric tenant includes, while ADLS Gen2 is storage you provision and manage yourself. OneLake shortcuts can point at ADLS Gen2, so many teams use both together.
Is Microsoft Fabric a data lake? Fabric includes a data lake and much more. OneLake is its lake, and Fabric adds data integration, engineering, warehousing, real-time analytics, data science and Power BI on top. OneLake is the shared storage layer. The compute engines and reporting tools all run against it, which is why one tenant can serve many workloads.
What is replacing Azure Synapse? Microsoft positions Fabric as the destination for analytics workloads that run on Azure Synapse today. Microsoft Learn publishes migration guidance for Synapse Spark workloads, covering notebooks, pipelines, lakehouse data and OneLake shortcuts. Check Microsoft Lifecycle for the support dates of the exact Synapse components you use before you set a timeline.
