Such as the title, I deploy a Sharded Cluster, in order to reduce the write pressure, so I use hash based sharding. But it will insert records randomly, is there anyways to decide it by myself?

Perhaps, if I can devide all sharding(e.g. : 10 shards) into two groups(group A <4 shards>, group B<6 shards>), A just insert users information, B just insert other information.