I want to store some time series data from several customers on an InfluxDb host. Due to privacy reasons, each customer must have its own database. But customers store only a few data, some only 5MB, others like 100MB.
Does the number of databases on a single host affect the performance significantly? I expect to have up to 500 databases on a single host. Or does the number not matter, and what matters is the sum of all data and requests across all databases?
Hi there,
InfluxDB is not technically designed to hold lots of databases in a single InfluxDB instance; rather it’s optimized to hold a high volume of series (data). Granted, I’m not certain what exact upper bound you would hit before your performance troubles would begin, so it’s possible having max 500 databases could be doable for your case.
Another option to consider is that fine-grained authorization/permissions capabilities are a part of InfluxEnterprise. You could store all customer data in a single database, and limit permissions via tagging.
Good luck!
Hi @Tkramm, I’m interested in following on this thread. By looking at the definition of ‘database’ in the influxDB glossary, it’s a logical container. (which shall be meaningful to represent the data if I’m not understand it wrong) In our case, we have separated contents sent by customers. Although it’s unusual to have such design to have multiple databases, it would be meaningful for us to organize the data. Your case sounds similar to us. Could you share a bit of your experience after applying such design? Do you run into any issue afterward? (e.g, performance) Thanks.
Other than that, is that still the case that a single InfluxDB instance is not designed to hold lots of databases?
