Hello,
I am in the process of researching and evaluating different tools for a custom logging service that we will use.
Our application is on premise for a number of clients, the app code + db is separate for each client. We want to move out of the main db some of the histories logging because it’s getting quite large. I wonder if TimescaleDB is suitable for such a task, with the following requirements:
- Use a single instance of TimescaleDB for all clients
- Each client must have its data in separate hypertable(s) from the other clients
- There are multiple types of logs
- There are different retention policies based on the type of the log and the client.
Because of the last point I am thinking to create a hypertable for each client and log type, thus retention policy per client and log type is easily implemented. The downside is that for each client I have to create about 10 hypertables. Number of clients is currently about 70 and maximum number of clients about 1000 tops. So 10000 hypertables.
My questions:
- Is such a structure with thousands of hypertables problematic for TimescaleDB?
- I would use 1 hypertable / 1 client, but I need to have different retention policy per log type as well.
- Would you say this is not a task for TimescaleDB but something else?
Many thanks,
Sorin