REST API vs FBDI for mass schedule imports: reliability, error handling, and performance comparison

ERP Oracle Fusion Cloud
, , , , , , , ,
1

We’re designing a schedule data integration that needs to import 5,000-10,000 schedule records daily from our planning system into Oracle Fusion Cloud. I’m evaluating REST API versus FBDI approaches and would appreciate insights from those who’ve implemented similar bulk imports.

REST API seems attractive for real-time validation and immediate feedback, but I’m concerned about rate limits and the overhead of individual API calls. FBDI appears better suited for bulk operations but error handling seems less granular - you get success/failure at the file level rather than record level.

Has anyone done a detailed comparison of these approaches for high-volume schedule imports? Particularly interested in understanding the trade-offs around bulk vs real-time integration patterns, how error handling differs in practice, performance characteristics, and whether hybrid patterns make sense.

2

We initially tried REST API for schedule imports but hit rate limits around 3,000 records. Switched to FBDI and processing time dropped from 45 minutes to 8 minutes for 10,000 records. The error handling in FBDI is actually record-level if you parse the output file correctly - it provides detailed error messages for each failed row.

3

Here’s concrete performance data from our implementation: REST API averages 200 records/minute (with proper batching), FBDI averages 1,500 records/minute. But REST API has predictable timing while FBDI varies based on server load. During peak hours, FBDI jobs can queue for 30+ minutes before processing starts. If you need guaranteed processing windows, REST API is more reliable despite being slower.

4

Error handling differs significantly. REST API gives you synchronous validation - you know immediately if a record fails and why. With FBDI, it’s asynchronous: submit file, wait for processing (usually 5-15 minutes for large files), download error report, parse failures. We built a monitoring framework that polls the FBDI job status and automatically processes error files. For critical failures, we have a fallback to REST API to retry individual records. Rate limits aren’t usually an issue if you batch REST calls appropriately - we do 50 records per call.

5

That hybrid approach is interesting. How do you handle the complexity of maintaining two integration paths? Also, what about error handling - with REST API you get immediate validation, but with FBDI you need to wait for the import process to complete and then parse results. How do you manage error notifications and retry logic in both scenarios?

6

Let me synthesize the key considerations based on extensive experience with both approaches:

Bulk vs Real-time Integration Patterns: FBDI excels at bulk operations (5,000+ records) with predictable schedules. It’s optimized for throughput, not latency. REST API shines for real-time or near-real-time scenarios where immediate feedback is critical. The performance crossover point is around 500-1,000 records - below this, REST API is competitive; above this, FBDI becomes significantly faster.

Error Handling Differences: REST API provides synchronous validation with immediate error responses. You can implement sophisticated retry logic, partial failure handling, and real-time error notifications. FBDI offers asynchronous processing with batch error reporting. The error file provides record-level detail (row number, field name, error message), but you need to build infrastructure to monitor job completion, retrieve error files, and process failures. REST API is better for interactive scenarios; FBDI works well for scheduled processes where immediate feedback isn’t required.

Performance and Rate Limits: REST API rate limits vary by service and license tier, typically 500-2,000 requests per hour. With batching (multiple records per request), you can process 10,000-50,000 records per hour. FBDI has no explicit rate limits but queuing delays during peak hours can add 15-45 minutes to processing time. For your 5,000-10,000 daily records, FBDI will consistently complete in 10-20 minutes; REST API could take 30-60 minutes depending on rate limits and batching efficiency.

Hybrid Integration Patterns: The most robust implementations use hybrid approaches: FBDI for scheduled bulk loads (daily imports, mass updates, initial data migration), REST API for event-driven updates (user-initiated changes, real-time corrections, critical priority records). This requires maintaining two code paths but provides optimal performance and flexibility. Implement a routing layer that directs records to the appropriate integration method based on volume, urgency, and business rules.

Practical Recommendations: For your 5,000-10,000 daily schedule imports, start with FBDI as the primary method. Build a monitoring framework that tracks job submission, polls for completion, retrieves and parses error files, and logs all failures to a database. Implement automatic retry for transient errors (use REST API for individual record retries). Add REST API as a secondary path for urgent updates or manual corrections. This hybrid approach gives you the performance of FBDI with the flexibility of REST API when needed.

The key is understanding that these aren’t competing technologies - they’re complementary tools for different integration scenarios.