Our senior DBA asked about Oracle's block size and I/O chunk size settings... are there optimal settings for the two values when running over NFS to a filer (a pair of F740's in our case)? Our Oracle instances run on Solaris 2.6, talking NFSv3 over UDP to the Netapps with 32K block sizes. My first reaction was to say "you don't need to tune this", but what do people use? It sounds natural to me to use 8K block sizes and an I/O chunk size of 4 to take advantage of the 32K NFS reads and writes. Does it make a difference?