I haven't used GridDB specifically, but a potential approach could be to implement a hybrid push-pull architecture to reduce the need for constant polling.

Since GridDB doesn't support native pub/sub or event triggers, you might consider introducing a lightweight intermediary such as Redis Pub/Sub or Kafka.

For example:

•    When new data is inserted into GridDB (via a data pipeline or ingestion service), you also publish a notification or payload to Redis or Kafka.

•    Your Streamlit app can then subscribe to these messages and update the UI accordingly (e.g., by triggering st.experimental_rerun() or updating cached data).

This setup avoids blind polling by having a dedicated mechanism to detect and propagate new data events to the dashboard.