Micro-Integration Flows now support configurable error handling to prevent data loss when messages fail during processing or delivery. Two independent features are available:

• Retry automatically re-attempts failed publishes using exponential backoff.
• Flow Dead Message Queue (DMQ) captures messages that can't be processed or delivered so you can inspect, reprocess, or discard them without losing data.

Both are disabled by default and can be enabled in any combination per Flow.

The error handling logic distinguishes between two failure types:

• Processing errors—such as validation failures or type mismatches—are considered unrecoverable and route directly to the DMQ without retrying, because retrying the same malformed message would always fail.
• Publishing errors—such as network issues or a temporarily unavailable target—are treated as potentially transient, so retry triggers first, with the DMQ as a fallback after retries are exhausted.

This feature is designed for sources that lack built-in error handling, like Change Data Capture (CDC) and polling-based sources. If your source system already has its own error handling mechanism, we recommend that you choose between using the source system's error handling or the Micro-Integration error handling—avoid enabling both at the same time. You configure error handling in the Error Handling step when you create or edit a Flow in Solace Cloud.

Performance improvements through architectural optimizations in message processing. These enhancements enable substantially higher message throughput and more efficient resource utilization, allowing for handling of larger workloads with the same infrastructure. In addition, further optimizations were done by enabling asynchronous publishing by default on Micro Integrations that support it, and disabling the default internal back-pressure for faster asynchronous publishing.

Various string and number functions are available to transform values to the desired target format, for example join, mask, abs, and many more.

This release supports JSON payloads only. Future releases will add other content formats such as XML. If your source and target payload is not JSON, you cannot map or transform the payload and it is passed through as is. However, you can still do header transforms and smart topic dynamic producer destinations.

This is a new capability and is not compatible with the "transform-headers" and "transform-payload" sections of the configuration available prior to this release. The "transform-headers" and "transform-payload" sections continue to work but are now deprecated and will be removed in a future release.

Existing connector configurations that rely on the legacy transform sections should not use the new transform capabilities. The legacy and new transform sections cannot be used together.

Note: This version supports IBM MQ Client 9.4.1.1.