What's New in Version 0.9.14

This section contains information about the new features and changes in the United Manufacturing Hub introduced in version 0.9.14.

Welcome to United Manufacturing Hub version 0.9.14! In this release we changed the Kafka broker from Apache Kafka to RedPanda, which is a Kafka-compatible event streaming platform. We also started migrating to a different kafka library in our micoservices, which will allow full ARM support in the future. Finally, we tweaked the overall resource usage of the United Manufacturing Hub to improve performance and efficiency, along with some bug fixes.

For a complete list of changes, refer to the release notes.


RedPanda is a Kafka-compatible event streaming platform. It is built with modern hardware in mind and utilizes multi-core CPUs efficiently, which can result in better performance compared to Kafka. RedPanda also offers lower latency and higher throughput, making it a better fit for real-time use cases in IIoT applications. Additionally, RedPanda has a simpler setup and management process compared to Kafka, which can save time and resources for development teams. Finally, RedPanda is fully compatible with Kafka’s API, allowing for a seamless transition for existing Kafka users.

Overall, Redpanda can provide improved performance and efficiency for IIoT applications that require real-time data processing and management with a lower setup and management cost.

Sarama Kafka Library

We started migrating our microservices to use the Sarama Kafka library. This library is written in Go and is fully compatible with RedPanda. This change will allow us to support ARM-based devices in the future, which will be useful for edge computing use cases. Addedd bonus is that Sarama is faster and requires less memory than the previous library.

For now we only migrated the following microservices:

  • barcodereader
  • kafka-init (used as an init container for components that communicate with Kafka)
  • mqtt-kafka-bridge

Resources tweaking

With this release we tweaked the resource requests of each default component of the United Manufacturing Hub to respect the minimum requirements of 4 cores and 8GB of RAM. This allowed us to increase the memory allocated for the MQTT broker, resulting in solving the common Out Of Memory issue that caused the broker to restart.

Be sure to follow the upgrade guide to adjust your resources accordingly.

The following table shows the new resource requests and limits when deploying the United Manufacturing Hub with the default configuration or with all the components enabled. CPU values are expressed in millicores and memory values are expressed in mebibytes.

CPU (default values)1080m (27%)1890m (47%)
Memory (default values)1650Mi (21%)2770Mi (35%)
CPU (all components)2002m (50%)2730m (68%)
Memory (all components)2873Mi (36%)3578Mi (45%)

The requested resources are the ones immediately allocated to the container when it starts, and the limits are the maximum amount of resources that the container can (but is not forced to) use. For more information about Kubernetes resources, refer to the official documentation.

Container registry

We moved our container registry from Docker Hub to GitHub Container Registry. This change won’t affect the way you deploy the United Manufacturing Hub, but it will allow us to better manage our container images and provide a better experience for our developers. For the time being, we will continue to publish our images to Docker Hub, but we will eventually deprecate the old registry.


  • Implemented a new test build to detect race conditions in the codebase. This will help us to improve the stability of the United Manufacturing Hub.
  • All our custom images now run as non-root by default, except for the ones that require root privileges.
  • The custom microservices now allow to change the type of Service used to expose them by setting serviceType field.
  • Added an SQL trigger function that deletes duplicate records from the statetable table after insertion.
  • Enhanced the environment variables validation in the codebase.
  • Added possibility to set the aggregation interval when calculating the throughput of an asset.
  • Various dependencies has been updated to their latest version.