We’re architecting a bidirectional EBOM synchronization between Aras 13.0 and our ERP system. The integration needs to handle complex BOMs with 5000+ line items and maintain real-time consistency between systems.
I’m evaluating REST API versus SOAP API for this integration. REST seems more modern and lightweight, but SOAP has built-in WS-Security and better transaction handling. The data mapping complexity is significant - we need to transform hierarchical BOM structures while preserving relationships and handling part substitutions.
Performance with large BOMs is critical since our manufacturing planning depends on up-to-date BOM data. Has anyone implemented large-scale EBOM sync using either API approach? What were the trade-offs in terms of performance, reliability, and handling complex data transformations? Particularly interested in experiences with batch operations versus incremental updates.
We implemented EBOM sync using REST API for Aras 13.0 to SAP integration. For large BOMs, we use batch operations with pagination - fetching 500 line items per request. REST performance is excellent, but you need to implement your own transaction handling and error recovery. We built a state machine that tracks sync progress and can resume from failures. The JSON payload for REST is more compact than SOAP XML, which helps with network transfer times. However, REST doesn’t have native support for distributed transactions, so you need to implement compensating transactions if the sync fails partway through.
Consider your ERP system’s capabilities too. If your ERP has better REST support, that might drive your decision. We sync BOMs between Aras and Oracle ERP using REST because Oracle’s REST APIs are more mature than their SOAP offerings. For performance with large BOMs, we implemented delta synchronization - only syncing changed items rather than full BOMs. This reduced sync times from 15 minutes to under 2 minutes for typical changes. The key is maintaining a change tracking mechanism in both systems so you know what needs to sync.
From a development perspective, REST API is much easier to work with. Modern tools and libraries have excellent REST support. For BOM data mapping, we use JSON transformations with JSONPath expressions to navigate hierarchical structures. Performance tuning is straightforward with REST - you can implement parallel requests for different BOM branches, implement caching strategies, and use HTTP/2 multiplexing. The challenge with REST is implementing reliable batch operations. We use a queue-based approach where BOM changes are queued and processed in batches every 5 minutes, with retry logic for failures.