Faster page loading

Ouva offers time-critical applications, and for some, every second counts. After listening to customer feedback, we drastically improved the patient page's loading time by adopting a layered approach to how we load data so that you can get the information as it appears instead of waiting for the entire page to load.

Fixes & improvements

  • Configured execution time metrics for all DB operations
  • Increased GraphQL response timeout to 30 seconds
  • Increased the number of application server threads to 16 to allow for more concurrent users.
  • Enabled gzip compression for all GraphQL and REST endpoints to minimize response payload.
  • Implemented asynchronous GraphQL resolvers to resolve a request concurrently.
  • Limited MongoDB pipeline execution timeout to 5 seconds.
  • Increased memory limits for some backend containers.
  • ChangeRoomState messages now contain bed coordinates to make segmentation work.
  • Added PATCH mapping for sensor update requests to avoid sending credentials during updates.
  • Fixed a bug causing some Kafka messages to get incorrect deserialization in Kafdrop.
  • Fixed a bug that causes the patient page to fire an additional query on load.
  • Speed up the attribute queue consumption by reading person crops from Redis instead of multiprocessing queue.
  • Fixed a bug related to sending pose messages to backend.
  • Improved room layout information usage by sending the initial coordinates at the start.