Hi, why you decide to build your own timeseries DB, and don’t use InfluxDB?
Chu Duc Minh
11

We tried it. It’s good, but it dos things we don’t want and that does slows it. Our one has a protocol buffers based wire protocol. So, it very fast on the network and very efficient. We also want to super fast and that’s why are build our own.

Our one currently have very predictable performance model and we process 200,000 writes/sec in a single node without any issues.