On 3/31/21 12:49 PM, Fred wrote:
Sorry for late reply. I think maybe I understand what you said. Do you mean that the file
system and erasure code of Ceph are ready enough for production, but it is strongly
depends on deployment and maintance correctly.
That's how I read it. You do have to take care how you build clusters:
type and amount of servers, network infra, etc. It all depends on the
amount of redundancy, performance you want to achieve, and how much
money you want to spend.
Besides your descripton, may I know where can I get
such data and evidence to enhance my confidence.
Best way is to find out for yourself and your use case(s) with a PoC
cluster. It's succesfully used in lots of places.