Home > Tutorials > Informatica to Fabric Migration

🟧 Tutorial 56: Informatica → Microsoft Fabric Migration¶

Last Updated: 2026-05-21 | Status: ✅ Final | Maintainer: Platform Team

📋 Table of Contents¶

• Overview
• Why migrate
• Component mapping
• Reference architecture
• Prerequisites
• Step-by-step migration
• Mapping translation patterns
• PowerExchange CDC migration
• License and cost analysis
• Validation checklist
• Troubleshooting
• References

📖 Overview¶

Informatica is a long-established, widely deployed ETL/data-integration platform with deep connectivity, mature data-quality and MDM capabilities, and a large skills base — genuine strengths that any migration should account for. The descriptions of Informatica products here are based on Informatica's publicly available documentation (as of this page's date); always verify against Informatica's current official docs. The two main product lines for migration source are:

1. PowerCenter — the classic on-prem ETL platform (mappings, workflows, sessions, repositories).
2. Intelligent Data Management Cloud (IDMC / formerly IICS) — the SaaS successor with Data Integration, Data Quality, MDM, B2B Gateway, and more.

This tutorial covers migrations from either product to Microsoft Fabric. The target is a Fabric F-SKU running Data Factory pipelines, Dataflows Gen2, PySpark notebooks, Mirroring, and Purview governance.

📝 Scope note: Informatica also has CDC (PowerExchange CDC), MDM, and a Data Quality product. Each gets its own dedicated section below since the target mappings differ.

🎯 Why migrate¶

The table below maps common drivers for choosing Fabric to the Fabric capability that addresses them. It reflects Informatica behavior as described in Informatica's publicly available documentation; where Informatica is the better fit for a given need, stay on Informatica. This is a comparison from a Microsoft Fabric perspective, not an authoritative assessment of Informatica.

🧭 Component mapping¶

The translation matrix below is based on Informatica's publicly documented component model (as of this page's date). Verify specifics against Informatica's current official documentation:

🏗️ Reference architecture¶

flowchart LR
    subgraph Informatica["🟧 Informatica (source)"]
        PC[PowerCenter<br/>repository]
        IDMC[IDMC / IICS<br/>SaaS]
        SA[Secure Agent<br/>on-prem connector]
        PXCDC[PowerExchange CDC]
    end

    subgraph Bridge["🔄 Migration bridge"]
        XMLEXP[XML export of<br/>mappings + workflows]
        SQLEXTRACT[Pre-existing SQL<br/>Source Qualifier extracts]
        SHIR[Self-hosted IR<br/>new connectivity]
    end

    subgraph Fabric["🟦 Microsoft Fabric (target)"]
        DF[Data Factory<br/>pipelines]
        DFG[Dataflow Gen2<br/>visual mappings]
        NB[PySpark notebooks<br/>code-first mappings]
        MIR[Mirroring<br/>CDC replacement]
        OL[(OneLake<br/>Bronze · Silver · Gold)]
        PBI[Power BI<br/>Direct Lake]
        PV[Microsoft Purview<br/>metadata + lineage]
        GE[Great Expectations<br/>+ Purview DQ rules]
    end

    PC --> XMLEXP --> DF
    PC --> XMLEXP --> DFG
    PC --> XMLEXP --> NB
    PXCDC --> MIR
    IDMC --> DF
    IDMC --> NB
    SA --> SHIR
    OL --> PBI
    Fabric --> PV
    Fabric --> GE

📋 Prerequisites¶

• ✅ Fabric F64 capacity with workspace identity (Tutorial 00)
• ✅ Self-hosted IR installed if any source is on-prem (Tutorial 23)
• ✅ Read-only access to the Informatica repository (PowerCenter) or IDMC org (IICS)
• ✅ XML export of all mappings and workflows from PowerCenter (pmrep command-line tool)
• ✅ Migration plan from Tutorial 13 signed off

🚀 Step-by-step migration¶

Step 1 — Inventory the Informatica estate¶

For PowerCenter, the pmrep CLI exports the full repository:

# Export all workflows from a folder
pmrep connect -r MY_REPO -d MY_DOMAIN -n admin -x <password>
pmrep listobjects -o workflow -f MY_FOLDER > workflows.txt
pmrep listobjects -o mapping  -f MY_FOLDER > mappings.txt
pmrep listobjects -o session  -f MY_FOLDER > sessions.txt

# Export to XML for analysis
pmrep objectexport -n MY_FOLDER -m -u workflows.xml

For IDMC, export via the Asset Management API:

curl -X POST "https://${POD}.informaticacloud.com/saas/api/v2/user/login" \
  -d '{"username":"…","password":"…"}' > login.json

SID=$(jq -r .icSessionId login.json)

curl -H "icSessionId: $SID" \
  "https://${POD}.informaticacloud.com/saas/api/v2/objects?type=mapping" \
  > idmc_mappings.json

Save the export — it drives the wave plan.

Step 2 — Bucket mappings by complexity¶

For each mapping, rate:

Step 3 — Pick a translation strategy per mapping¶

Three strategies, by mapping shape:

3a. Visual: Dataflow Gen2 (preferred for simple/medium)¶

Use when the mapping is dominantly Source → Transform → Sink. Dataflow Gen2 has a Mapping Editor very similar to PowerCenter's, with Power Query M behind the scenes.

3b. Code: PySpark notebook (preferred for complex)¶

Use when you have Java/Python transformations, dynamic schemas, or custom SQL overrides. The PySpark equivalent is usually shorter than the original Informatica mapping.

3c. SQL: Fabric Warehouse stored proc¶

Use when the mapping is basically INSERT INTO target SELECT ... FROM source and the source is already in Fabric Warehouse.

Step 4 — Land source data in Bronze¶

For each Informatica source, decide:

Step 5 — Port mappings (wave by wave)¶

Convert each mapping to its Fabric equivalent. See Mapping translation patterns below for line-by-line examples.

Step 6 — Re-orchestrate workflows as Fabric pipelines¶

Each PowerCenter workflow / IDMC taskflow becomes a Fabric Data Factory pipeline. Sessions become Dataflow Gen2 / notebook activities. Decision tasks become If Condition activities. Event-wait tasks become Wait activities or webhook-triggered pipelines.

Step 7 — Migrate Data Quality rules¶

Informatica DQ rules → Great Expectations suites. Pattern:

# Great Expectations suite mirroring an Informatica DQ rule
import great_expectations as ge

df = spark.read.table("lh_silver.silver_customer")
ge_df = ge.from_spark(df)

# Original Informatica rule: "customer_id must be unique"
ge_df.expect_column_values_to_be_unique("customer_id")

# Original: "email must match regex"
ge_df.expect_column_values_to_match_regex("email", r"[^@]+@[^@]+\.[^@]+")

Purview classification rules pick up the rest (PII detection, glossary mapping).

Step 8 — Cutover¶

1. Run Informatica and Fabric in parallel for one full load cycle.
2. Reconcile row counts and key sums for every target table.
3. Repoint downstream consumers (Power BI, ML, external apps).
4. Disable the Informatica session schedule.
5. Archive the Informatica repository XML to lh_archive.
6. Cancel the Informatica subscription at renewal.

🧩 Mapping translation patterns¶

Concrete pattern-by-pattern translation between Informatica transformations and Fabric equivalents.

Lookup (cached) → PySpark broadcast join¶

Informatica: Lookup transformation with the "Cache" option enabled.

Fabric (PySpark):

from pyspark.sql.functions import broadcast

dim_customer = spark.read.table("lh_silver.dim_customer")     # small
fact_sales   = spark.read.table("lh_silver.fact_sales")        # large

joined = fact_sales.join(broadcast(dim_customer), "customer_id", "left")

Update Strategy (DD_UPDATE / DD_INSERT) → Delta MERGE¶

Informatica: Update Strategy transformation with IIF(condition, DD_UPDATE, DD_INSERT).

Fabric (PySpark SQL):

spark.sql("""
  MERGE INTO lh_silver.silver_customer t
  USING staging_customer s
    ON t.customer_id = s.customer_id
  WHEN MATCHED THEN UPDATE SET *
  WHEN NOT MATCHED THEN INSERT *
""")

Aggregator → groupBy().agg()¶

Informatica: Aggregator with SUM(amount) grouped by customer_id.

Fabric:

from pyspark.sql.functions import sum as _sum

result = df.groupBy("customer_id").agg(_sum("amount").alias("total_amount"))

Router → multi-branch filter + union¶

Informatica: Router with 3 output groups (Active, Inactive, Pending).

Fabric:

active   = df.filter(df.status == "ACTIVE")
inactive = df.filter(df.status == "INACTIVE")
pending  = df.filter(df.status == "PENDING")

# Write each to its own target
active.write.saveAsTable("lh_silver.silver_active_cust")
inactive.write.saveAsTable("lh_silver.silver_inactive_cust")
pending.write.saveAsTable("lh_silver.silver_pending_cust")

Sequence Generator → Delta identity column¶

Informatica: Sequence Generator with a Reusable shared sequence.

Fabric (Delta Lake):

CREATE TABLE lh_silver.silver_customer (
  customer_sk BIGINT GENERATED ALWAYS AS IDENTITY,
  customer_id STRING,
  -- other columns
) USING delta;

Stored Procedure transformation → Pipeline Stored Proc activity¶

Drop into a Fabric Warehouse stored procedure activity. Pass parameters via pipeline expressions.

🔄 PowerExchange CDC migration¶

PowerExchange CDC is one of Informatica's most-licensed components. The Fabric equivalents:

For each PowerExchange CDC instance:

1. Identify the source DB type.
2. Pick the Fabric pattern from the table above.
3. Stand up the equivalent in parallel.
4. Reconcile CDC events for 24-48 hours.
5. Cut over by disabling the PowerExchange capture.

See Mirroring for the configuration details.

💰 License and cost analysis¶

Cost outcomes vary widely by estate, contract terms, and workload mix, so treat any figures here as illustrative only — build your own model from your actual Informatica contract and a measured Fabric capacity sizing. Some organizations see meaningful steady-state savings; others, depending on their licensing and usage, see less. Note too that the migration year often costs more than simply continuing on Informatica because both platforms run in parallel — plan for it.

Informatica licensing is negotiated per-customer and is not generally published; the ranges below are rough, illustrative placeholders only and are not sourced from Informatica's official pricing. Replace them with your own contracted numbers before making any decision.

Migration year additional costs:

• Parallel run: 6 months at 2× compute
• Migration partner (if used): \(500K-\)2M for a medium estate
• Internal labor: 2-4 engineers × 6 months × loaded cost

✅ Validation checklist¶

• Every Informatica source connected via the Fabric equivalent
• Every mapping ported (Dataflow Gen2, notebook, or pipeline activity)
• Every workflow re-orchestrated as a Fabric pipeline
• PowerExchange CDC replaced by Mirroring or equivalent
• DQ rules ported to Great Expectations / Purview
• MDM (if any) rebuilt in Fabric IQ
• Row counts and key sums reconcile between platforms (≤ 0.5% drift)
• Downstream consumers (PBI, ML, APIs) repointed
• Final repository XML archived to lh_archive
• Informatica subscriptions cancelled at renewal

🛠️ Troubleshooting¶

📚 References¶

• Tutorial 13 — Migration Planning
• Tutorial 23 — SHIR & Data Gateways
• Tutorial 25 — IBM DB2 Source
• Tutorial 26 — Multi-Source Streaming
• Mirroring
• Copy Job CDC
• Dataflow Gen2
• Best practices — Migration patterns
• Validation framework
• Fabric IQ — MDM replacement
• Purview — metadata manager replacement

Navigation: ⬅️ 55 — Palantir → Fabric | Tutorials Home | Migration Planning ➡️