On Mon, Dec 21, 2009 at 2:37 AM, <[email protected]> wrote: > 1. How does SStables of one key-space distribute among multiple data > directory?(what distribute policy?)
round-robin > 2. What fault-tolerant policy does cassandra utilize to against disk > failure? node-level replication > 3. If one disk down, What impact does one key-space have on reading > and writing?(does whole key-space become unavailable?) no, but you should run nodeprobe repair to re-replicate the data from the dead disk to good ones. -Jonathan
