Hi everyone,
I’m currently working on a project that involves storing and analyzing large volumes of time-series sensor data in GridDB. I’ve been exploring the documentation and experimenting with some sample use cases, but I’d appreciate some guidance from the community on writing efficient queries for real-time and historical analytics.
My use case involves inserting thousands of records per minute from multiple IoT devices (temperature, humidity, pressure sensors). Each record is timestamped, and my primary need is to run frequent queries that fetch data over specific time ranges (e.g., last 15 minutes, last 24 hours) or aggregate values like averages, max/min, etc.
So far, I’ve implemented containers for each sensor type and structured them using time-series containers. However, I’m unsure if I’m optimizing the queries well. Specifically, I’m looking for best practices around:
- Querying time-series containers using TQL — is there an efficient way to filter by timestamp and device ID?
- Using indexes — what are the best indexing strategies to improve query performance on time fields?
- Aggregation performance — how well does GridDB handle aggregation functions on large datasets?
- General tuning — are there any tuning tips (buffer sizes, partitioning, etc.) that can improve overall performance for time-series workloads?
Any shared experiences, example queries, or performance benchmarks would be incredibly helpful. If there’s a better modeling pattern for sensor data of azure training in pune, I’m open to redesigning early.
Thanks in advance!