Revisited from the Foresaw archive. The argument and structure have been updated for this edition.
Moving ERP integration into the cloud can change how teams provision, connect, and operate systems. It does not automatically make those systems agree about the business.
The useful question is not whether the integration is cloud-based. It is whether the architecture makes data movement, failure, and ownership easier to manage.
Separate deployment from design
Managed infrastructure can remove some operational work, but the team still needs to define the integration contract. What does each record mean? Which system owns a change? How long can another system use an older value?
Answer those questions independently of the hosting choice. They determine the behavior people will experience when the system is under pressure.
Match capacity to the workload
A periodic import, a burst of orders, and an interactive lookup behave differently. Model the expected volume and timing before choosing how the integration scales.
Include the limits of the ERP and connected applications. Increasing the capacity of a worker does not increase the number of requests a downstream system can accept.
Keep costs and failures visible
Understand what drives usage: data transfer, execution time, storage, and repeated retries. Observe the complete transaction across service boundaries so that a failure does not disappear between providers.
Assign operational responsibility
Know who responds when an integration stops, who can replay work, and how reconciliation is performed. Security configuration, access permissions, and retention settings also need owners. A cloud platform can support the design; it cannot make those decisions for the business.
Adapted from “The Role of Cloud Technology in ERP Integrations.”
First published on Foresaw. Reworked for this edition.