On Wed, 2015-12-30 at 22:10 +0800, Qu Wenruo wrote: > Or, just don't touch it until there is really enough user demand. I definitely think that there is demand... as I've written previously, when I did some benchmarking tests (though on MD and HW raid) then depending on the RAID chunk size, one got different kind of IO patterns tuned.
> (In stripe_len case, it's also a possible choice, as configurable > stripe > length doesn't really affect much except RAID5/6) Sure about that? Admittedly I haven't checked it for the not parity RAIDs, but I'd expect that for the same reasons you get out different performance for sequential/random/vector reads/writes, you'd also get that more or less at least at RAID0. Cheers, Chris.
smime.p7s
Description: S/MIME cryptographic signature
