Integrationintermediate
HCM Data Loader (HDL)
Explain how HDL is used for bulk loading and updating HCM data using a flat-file, delta-friendly format
HDL is HCM's bulk loader. It exists separately from FBDI because HCM data is deeply hierarchical in a way Excel templates handle badly.
HDL is like a detailed family-tree import file for a genealogy website: rather than uploading disconnected individual records, you submit one structured file expressing exactly how each person (worker) relates to their associated records (assignments, addresses), and the system intelligently merges new information into existing family trees or creates new branches as needed.
Key Concepts
1
A worker is not one record.
Person
Work Relationship
Assignment
Salary
Element Entries2
Loading a new hire means all of those, in order, correctly linked. An Excel template per object would be
five files with keys that must match.
3
HDL uses a DAT file expressing the hierarchy in one place.
METADATA|Worker|PersonNumber|StartDate|...
MERGE|Worker|12345|2026-01-15|...
METADATA|WorkRelationship|PersonNumber|LegalEmployerName|...
MERGE|WorkRelationship|12345|Acme India|...
METADATA|Assignment|PersonNumber|AssignmentNumber|JobCode|...
MERGE|Assignment|12345|E12345|SWENG|...4
METADATA declares the columns for the object that follows. MERGE is a data line.
METADATAMERGE
5
Several objects in one file, loaded together as a unit — which is what the hierarchy requires.
6
MERGE means insert-or-update, decided per record. That is what makes HDL usable for recurring feeds:
MERGE
7
a new person -> inserted
an existing person -> updated with whatever attributes you sent8
One file handles both. There is no separate insert and update path, and no need to know in advance which
records already exist.
9
Delta loading follows from that. Send only the attributes that changed for the people who changed.
Unsent attributes are left alone.
10
That is exactly what a nightly sync from an external HR system needs — and it is why HDL suits that job
better than a full extract-and-replace.
11
Uploaded to UCM, like FBDI, and processed by the Load and Import job. The logs identify per-record
outcomes with messages traceable to the source line, so a handful of bad records do not block the rest.
12
The scenario to be ready for. Loading a new hire complete — person, work relationship, assignment,
salary — in one coordinated file, and explaining why the order and the linking keys matter.
13
And the caution. MERGE on a record you did not intend to change will change it. A feed sending a field
it should not own overwrites what HR entered, quietly, every night. Deciding which system owns which
attribute is the design work; HDL just does what the file says.
MERGE
14
What the interviewer is probing.1. "What is HDL for?" Probing: the purpose. Stalls: "Loading data." Moves up: bulk loading
HCM business objects from pipe-delimited .dat files, which is the supported path for conversions and
ongoing interfaces.
15
2. "What makes HDL files tricky?" Probing: the real difficulty. Stalls: "The format." *Moves
up:* date-effective records and the parent-child hierarchy — a worker needs person, assignment and
work relationship rows consistent on dates.
16
3. "How do you correct a bad load?" Probing: the practical skill. Stalls: "Delete the rows."
Moves up: reload with corrected data using the same source keys, or use the delete action — direct
data fixes are not available.
17
4. "What is HSDL and when would you use it?" Probing: the breadth. Stalls: "The same."
Moves up: HCM spreadsheet data loader, for smaller business-user-driven loads where a full .dat
file is overkill.