Batch approval optimization for faster engineering change processing in approval-mgmt module (env-r2022x)

PLM ENOVIA
, , , , , , , ,
1

We recently optimized our batch approval process for engineering changes and achieved significant performance improvements. Our ECO approval cycles were taking 45+ minutes for batches of 200+ changes, creating bottlenecks during peak periods.

The solution involved three key improvements: implementing parallel processing for approval routing, optimizing batch notification generation, and adding real-time workflow monitoring.

Here’s the core parallel processing implementation:

ExecutorService executor = Executors.newFixedThreadPool(10);
for (ChangeRequest2 eco : ecoBatch) {
    executor.submit(() -> processApproval(eco));
}
executor.shutdown();

After optimization, batch processing time dropped to 12 minutes - a 73% improvement. Notification generation was reduced from 15 minutes to 3 minutes through batching and template caching. We also implemented workflow monitoring dashboards to track approval progress in real-time.

Happy to share more details about configuration changes and monitoring setup.

2

For notifications, we implemented two-level optimization. First, template caching using Guava cache with 30-minute TTL - eliminates repetitive template queries. Second, batch database operations using JDBC batch inserts for notification records instead of individual saves. This reduced database roundtrips by 85%.

For monitoring, we built custom MBeans exposing metrics like approval queue depth, processing rate, and average completion time. These feed into Grafana dashboards. The real-time visibility helps us spot bottlenecks immediately and adjust thread pools dynamically during high-load periods.

3

Did you encounter any transaction management issues with parallel processing? Concerned about potential deadlocks or consistency problems when multiple threads process approvals simultaneously.