Notes
Workday integration notes
Short answers to questions I get asked about Workday integration work.
- 01What does a Workday integration consultant do?
- 02What is the difference between Workday Studio, Extend and Orchestrations?
- 03When should you use Message Queue Service instead of a scheduled integration?
- 04Why does at-least-once delivery mean consumers must be idempotent?
- 05Why does XSLT 3.0 streaming matter for large payroll files?
- 06What makes this Workday integration work different, and where has it been done?
What does a Workday integration consultant do?
A Workday integration consultant designs, builds and maintains the data flows between Workday and other systems: payroll providers, benefits carriers, banks, ServiceNow and finance platforms. The work spans Workday Studio, Core Connectors, EIBs, RaaS, PECI/PICOF, Workday Extend and Orchestrations, plus the security configuration and business processes that keep those integrations running. The job is as much about the edges as the happy path: what happens when a file is late, a downstream system is down, or the same event arrives twice.
What is the difference between Workday Studio, Extend and Orchestrations?
Workday Studio is the integration toolset for moving and transforming data between Workday and external systems, using XSLT, web services and mediation patterns. Workday Extend is a platform for building custom applications inside Workday, with custom business objects, REST APIs, role-based security and automated workflow steps. Orchestrations are low-code, event-driven automations on the Extend platform. They trigger on Workday events and chain steps across systems, for example syncing a worker's updated job title from Workday to ServiceNow with no scheduled batch job. As a rule of thumb: Studio integrates, Extend builds net-new apps, and Orchestrations connect events to actions.
Longer answer: Studio vs Extend vs Orchestrations→
When should you use Message Queue Service instead of a scheduled integration?
Scheduled polling (EIB or RaaS on an interval) works well when near-real-time delivery is not required and downstream systems are reliably available. It breaks down when a downstream system has intermittent availability and events cannot wait for the next batch window. Workday Message Queue Service enqueues each event and retries delivery until it is acknowledged, regardless of whether the downstream system was up when the event occurred. Paired with Studio Listener Service for event-driven inbound triggers, it replaces clock-driven integration with event-driven delivery. The tradeoff moves to the receiving side: delivery is at-least-once, never exactly-once, so consumers must be idempotent and safely handle the same message twice.
Longer answer: MQS vs scheduled integrations→
Why does at-least-once delivery mean consumers must be idempotent?
A queue that retries until it gets an acknowledgement will occasionally deliver a message more than once, for instance when the consumer processed it but the acknowledgement was lost. Retrying is what prevents lost messages, so duplicates are the price of that guarantee. Message Queue Service therefore delivers at-least-once, not exactly-once. The fix belongs in the consumer: make processing idempotent, so handling the same message a second time produces the same result as handling it once. Typical approaches are keying each operation on a stable event identifier, using upserts instead of blind inserts, and recording which identifiers have already been applied. You can try the retry and outage behaviour in the MQS Simulator among the free tools.
Why does XSLT 3.0 streaming matter for large payroll files?
Traditional XSLT builds the entire source document as an in-memory tree before producing output. That becomes a real constraint on large payroll extracts inside Workday Studio's resource-governed runtime. XSLT 3.0 streaming reads, transforms and emits output node by node without holding the full tree in memory, so memory use stays roughly flat regardless of file size. It requires designing the transform to be genuinely streamable: single forward-pass grouping with group-starting-with, and avoiding constructs that need random access to the tree. The payoff is reliable extracts at a scale where conventional processing is not viable, as built for a payroll extract covering 30,000+ employees at Manulife.
What makes this Workday integration work different, and where has it been done?
The emphasis is on the advanced side of the platform: production XSLT 3.0 streaming for files beyond memory limits, Message Queue Service for reliable asynchronous delivery, and Studio Listener Service for event-driven integration. That sits on 25+ years in software, including 14 years of engineering from 2000 to 2014 before moving into Workday in 2014, and 12+ years in Workday since. Nine certifications are active, eight Workday and one MuleSoft. The work spans 10+ industries, including financial services, higher education, government, power and utilities, insurance, retail, mining, health, and hiring and background checks. I'm based in Canada and available for contract and advisory engagements, working remotely with teams across North America.