Microsoft Fabric, Data Foundation and Training
Microsoft Fabric Consulting, Implementation and Training
Capacity sized before it throttles, OneLake shortcuts instead of copies, Power BI models that read the lake directly, a migration path off Synapse and Premium, and Purview labels that follow the data.
Microsoft Fabric consulting is help planning and running Fabric well: sizing and sharing capacity, using OneLake shortcuts rather than copying data, building Power BI semantic models on Direct Lake, moving off Synapse and Power BI Premium, and connecting Microsoft Purview. Kriv AI offers it as a Microsoft "Solutions Partner for Cloud & AI Platforms · Partner ID 7036589". We hold no Fabric specialization and have no named Fabric case study yet.
capacity planning
Microsoft Fabric Consulting Starts With Capacity
Fabric is Microsoft's software-as-a-service analytics platform: data engineering, warehousing, data science, real-time workloads and Power BI in one service, on one shared store called OneLake. Almost everything in it draws on a purchased capacity, which is why capacity is the first design decision, not an afterthought.
A single capacity shared by every team is the most common early setup, and the most common source of complaints. A data engineer's backfill and the finance team's month-end reports draw on the same pool, and when the pool is overloaded everything slows down together. Our capacity work answers four questions: which workloads run where, how production and development are separated, who is allowed to assign workspaces to which capacity, and who watches utilization.
In practice that usually means separate capacity for production reporting and for engineering or development work, workspace assignment limited to administrators, and the capacity metrics reviewed on a schedule by a named owner. We size against your observed usage during a pilot rather than a guess, and revisit the sizing after the first migration wave lands.
onelake shortcuts
OneLake Shortcuts: One Copy of the Data
OneLake is the single logical data lake every Fabric workspace in a tenant shares. Shortcuts let a workspace reference data that lives elsewhere, in another workspace, in Azure Data Lake Storage, or in other supported cloud storage, without copying it.
Shortcut first, copy by exception
Our default rule is that a team needing another domain's table creates a shortcut to it rather than a pipeline that copies it. Copies are allowed only for a stated reason, such as a different retention requirement, and are recorded with their owner. This is what stops a shared lake from filling up with five versions of the same customer table.
Domains and a workspace naming standard
Workspaces are grouped into Fabric domains by business area, and named by domain, purpose and environment, for example a claims engineering workspace separate from claims reporting, each with development and production versions. Each domain has an owner who approves shortcuts into its data.
Access follows the source
Because a shortcut points at data rather than duplicating it, we check how permissions apply at both ends before a shortcut crosses a sensitivity boundary. That review is part of the shortcut request, not a later audit.
power bi semantic models
Power BI Consulting: Semantic Models and Direct Lake
Direct Lake is a Power BI storage mode in which a semantic model reads Delta tables in OneLake directly, without a scheduled import of the data. It changes how semantic models should be designed, and it is where Power BI consulting inside a Fabric program spends most of its time.
| Decision | Our default | Why |
|---|---|---|
| How many semantic models | A small set of shared, owner-approved models per domain; reports connect to them rather than carrying their own. | One definition of each measure, and one place to apply row-level security. |
| Which storage mode | Direct Lake where the gold tables are well shaped; import where a model needs logic the lake tables cannot provide. | Direct Lake works best on tables designed for it, so table design and model design are agreed together. |
| Consolidation criteria | Models with overlapping tables, the same measures under different names, or no recent use are merged or retired. | Fewer models to secure, document and migrate. |
| Who can publish | Shared models published only from the production workspace through a deployment pipeline; personal workspaces for drafts only. | A reported number can be traced to an approved model and its change history. |
fabric migration
Microsoft Fabric Migration from Synapse, Legacy Warehouses and Power BI Premium
Fabric migrations rarely have one source. Most of ours have three, each with its own path.
From Azure Synapse Analytics
Dedicated SQL pool schemas and T-SQL code move to a Fabric warehouse, with code reviewed for features that behave differently. Spark notebooks and pipelines move to Fabric data engineering items. Where the data already sits in Azure Data Lake Storage, shortcuts let Fabric read it during the transition, so reports can move before the storage does.
From on-premises or other legacy warehouses
SQL Server and other legacy warehouses are moved table group by table group, with ingestion rebuilt as Fabric pipelines or dataflows. Semantic models are re-pointed only after their tables match the legacy source on agreed reconciliation checks.
From Power BI Premium capacity
Moving workspaces from Power BI Premium capacity onto Fabric capacity is usually the simplest step technically, and the best moment to retire unused reports and consolidate duplicate semantic models before they are carried forward.
purview governance
Governance with Microsoft Purview
Fabric integrates with Microsoft Purview, Microsoft's family of data governance and compliance tools, for cataloguing, sensitivity labels and audit. If your organization already labels documents and email in Microsoft 365, Fabric can be brought under the same scheme.
We agree which sensitivity labels apply to Fabric items, which items inherit labels from their sources, and who may lower a label. We make sure Fabric activity reaches the audit log your security team already uses, and we decide which items the Purview catalog should surface, so people can find endorsed data before they build their own. Label and audit rules are not invented from scratch: we write the rules to the policy and frameworks you already answer to, such as HIPAA, 21 CFR Part 11, SR 26-2 or GDPR.
fabric training
Microsoft Fabric Training by Role
Our Microsoft Fabric training runs on your own tenant and data wherever possible, during the build, so people learn on the workspaces they will use.
1. Analysts and report authors
Finding endorsed data through the catalog, building reports on shared semantic models, what Direct Lake means for refresh and performance, and when to ask for a model change instead of building a private copy.
2. Data engineers
Lakehouse and warehouse items, pipelines and notebooks, shaping gold tables so Direct Lake models perform, creating shortcuts under the domain rules, and Git integration with deployment pipelines.
3. Fabric administrators
Tenant settings, capacity assignment and the capacity metrics, domain and workspace administration, and reviewing audit activity.
4. Data and domain owners
A short session on approving shortcuts and access, endorsing semantic models, applying sensitivity labels, and signing off migrated content.
implementation partner
Working With Us as Your Microsoft Fabric Implementation Partner
We start with a review of your tenant and Power BI estate: capacities, workspaces, semantic models, report usage and existing labels. From that we agree the order of work, which usually means capacity and domain design first, then one domain migrated end to end as the pattern for the rest. We build alongside your team and hand over each domain with an owner, a runbook and trained people.
Kriv AI holds Microsoft's "Solutions Partner for Cloud & AI Platforms · Partner ID 7036589". Partner Center groups three designations under it, "Data & AI", "Infrastructure" and "Digital & App Innovation", and our Data & AI designation was attained in Jul 2026. We are also a "Databricks partner", useful when both platforms are in play. We have no named Fabric engagement to publish yet; ask during discovery what we can share.
get started
Starting a Fabric Engagement
Fabric consulting and training are scoped and priced after a short discovery call. Bring a rough picture of your capacities, your Synapse or warehouse footprint and your Power BI estate; the call ends with a recommended first domain and first deliverable.
Straight answers
Frequently asked questions about Microsoft Fabric Consulting, Implementation and Training
How much Fabric capacity do we need?
It depends on your workloads, so we avoid sizing from a rule of thumb. We run a pilot on representative work, watch the capacity metrics, and size from what we observe, usually separating production reporting from engineering and development. We revisit sizing after the first migration wave, because real usage after migration rarely matches the estimate made before it.
How much does Microsoft Fabric consulting cost?
Microsoft Fabric consulting is scoped and priced after a short discovery call. What drives the size of the work is the number of capacities and workspaces, whether Synapse or a legacy warehouse is being migrated, how many semantic models need consolidating, and whether training is included. The pricing page describes our engagement models in general; the call produces a proposal for your tenant.
Should we use OneLake shortcuts or copy data into Fabric?
Shortcut by default. A shortcut references data where it already lives, so there is one copy to secure and keep current. Copy only for a stated reason, such as a different retention rule or a source that cannot be reached directly, and record who owns the copy. Before a shortcut crosses a sensitivity boundary, check how permissions apply at both ends.
Should our Power BI semantic models use Direct Lake?
Where the underlying lake tables are well shaped for reporting, Direct Lake removes the scheduled import and keeps reports closer to the data. Where a model needs logic the lake tables cannot provide, import mode can still be the right choice. We decide model by model, and we design the gold tables and the semantic model together so Direct Lake performs well.
Can you migrate us from Azure Synapse to Fabric?
Yes. Dedicated SQL pool code moves to a Fabric warehouse after review for features that behave differently, and Spark notebooks and pipelines move to Fabric data engineering items. Where data already sits in Azure Data Lake Storage, shortcuts let Fabric read it during the transition, so reporting can move before storage does. Each table is reconciled against the source before cutover.
Do you offer Microsoft Fabric training on its own?
Yes, though it works best during a build. We run tracks for analysts and report authors, data engineers, Fabric administrators, and data owners, on your own tenant and data where possible. If you only want training, we start with a short look at your capacities, domains and semantic models so the material matches how your Fabric estate is actually set up.
How does Microsoft Purview fit with Fabric?
Purview provides cataloguing, sensitivity labels and audit across Microsoft services, and Fabric integrates with it. We set which labels apply to Fabric items and which inherit from sources, make sure Fabric activity reaches your audit log, and decide what the catalog surfaces. For regulated teams we write these rules to the frameworks you already answer to, such as HIPAA.
Talk to the team that would do the work
Bring your requirements to a working session with the person who'll actually deliver.
Book a Discovery Call