ERP NOTE · MFG/A · SYTELINE · REV 2026-10
For the controller, the COO, and whoever owns the reports

Syteline in Power BI and Fabric:
reach the data, then define the number.

Syteline already holds what a margin, delivery, inventory, or capacity report needs. The reports disagree anyway, and the reasons are specific to how Syteline stores sites, dates, and costs. This note covers three things: how the data comes out, where the numbers split, and what gets defined first.

How the data comes out

It depends on where your Syteline runs. On premises and multi-tenant cloud have almost nothing in common, so settle which case you are in before anyone designs a report.

DeploymentWhat you can reachHow it lands in Fabric
On premises The SQL Server application database, with a read-only login. Through an on-premises data gateway, by mirroring, a copy job, or a pipeline.
Multi-tenant CloudSuite Industrial No SQL connection to the application database. Data leaves through Infor’s Data Lake. Out of the lake by Compass, Stream Pipelines, or the ETL client, then into Fabric.
Hosted, single tenant It depends on the contract. Some arrangements include database access and some do not. As on premises if you have database access. As the multi-tenant case if you do not.

On premises: read the base tables

  • Since version 9, most application tables carry an _mst suffix and a site_ref column. Customer order lines are coitem_mst, jobs are job_mst, material transactions are matltran_mst.
  • The same-named views without the suffix filter to a site that the application sets for its own sessions. An outside connection has not set it, so a report built on the views can come back empty. Read the _mst tables and filter site_ref yourself.
  • One database can hold several sites, and a SQL login sees every site’s rows. Every measure has to say whether it is one site’s number or the company’s.
  • Order history can sit in separate history tables, coh and citemh. A delivery measure that reads only the live order tables can lose its past.
  • Fabric mirroring for SQL Server keeps a continuous copy in OneLake. On SQL Server 2016 to 2022 it relies on change data capture and needs a primary key on every mirrored table. It copies tables, not views, and stops at 1,000 tables, so choose them on purpose.
  • Change data capture is a change on the ERP database server. Clear it with whoever supports your Syteline before turning it on. A scheduled copy job reads through an ordinary login and changes nothing on the server.

Multi-tenant cloud: start with what is in the lake

  • Syteline sends data to the Infor Data Lake, but only for tables in the Data Lake replication category. Not every table is there. What the lake holds is the first question, ahead of any tool.
  • Replication covers changes from the day it is switched on. Existing history has to be pushed with the Manual Replication Utility.
  • The lake keeps each change to a record as a new variation. Compass, the lake’s SQL engine, can return the latest variation and leave out deleted records. A raw object pull does not, and the extract has to do it.
  • Three routes lead out: Compass through its API or JDBC driver, Stream Pipelines into an Azure SQL database, or the Data Fabric ETL Client into a SQL Server you run. Which of these your tenant is licensed for is the second question.
  • Neither Power BI nor Fabric has a Microsoft-built connector for Infor. Whatever route you take, the data passes through something you configure and own.
  • The IDO REST API and ION documents are integration tools. They suit moving a transaction between systems, not feeding years of history to a model.

Where Syteline numbers split

The general causes are in Why your reports disagree. These are the forms they take in Syteline. Each row ends in a decision somebody has to make once and write down.

NumberWhat Syteline holdsThe decision to write down
On-time delivery An order line carries a due date, a request date, and a projected date. When planning re-dates a line, the original due date is kept as the request date. The actual ship date comes from the shipping transactions. Infor’s own delivered-in-full-and-on-time report measures against the due date, with tolerance days. Which date is the promise, and how many days early or late still count.
Margin Each item has a cost type, actual or standard, and a cost method: average, LIFO, FIFO, specific, or standard. Syteline keeps a current unit cost that moves with receipts and roll-ups, and a standard unit cost that holds until it is rolled. One cost basis for margin, stated by name.
Job cost Until a job closes, finished quantity can move to inventory at planned cost, actual cost, or a mix, depending on a costing parameter. Under standard cost, the difference goes to variance accounts at close. Whether job margin is read before or after close, and where variances belong.
Custom fields User extended tables add fields to standard tables, with names that begin Uf_. Logic written in IDO extension classes runs in the application layer, so a value it computes may never be stored in a table. Which custom fields carry meaning the reports need, and where each value is actually stored.

What gets defined first

  • Access. What can be reached today, by which route, and who controls it. The argument is in Access is the precondition.
  • Site and calendar. Which sites a number covers, and which fiscal calendar it follows.
  • On-time delivery. One date, one tolerance, one treatment of partial shipments.
  • Margin. One cost basis, with the job-close rule stated.

Then the build follows the same order every time. Extracts land untouched in a Fabric lakehouse in your tenant. They are conformed to those definitions. One semantic model sits on top, and every Power BI report reads it. Inventory, quality, and capacity are added to the same model, each defined once.

What the audit establishes on Syteline

  • Which deployment you are on, and exactly what can be reached from outside it.
  • Which sites share a database, and which reports mix them.
  • Which date and which cost basis each existing report uses today.
  • Which modules are healthy, misconfigured, or abandoned.

We also work on Global Shop Solutions.

Product names and limits are as Infor and Microsoft document them in October 2026. They change, and contracts differ. The audit checks each one against your own deployment.

Where this starts

The audit prices this before you spend a dollar.

Two weeks, $3,000, fixed scope. A graded report on how your ERP is actually used, where the numbers disagree, and a priced recommendation. You keep the report either way.

Book the ERP data audit