On Dec 16, 2007 2:26 PM, Andrew Lentvorski <[EMAIL PROTECTED]> wrote: > Bob La Quey wrote: > > > Because ZFS runs on the "main" processor not some peripheral > > processor on a disk controller it will cause a hit on > > the "main" processor ... but the advantage of that is the "main" > > processor is far more likely, because of economies of scale > > and competition, to be far faster than the cpus used in the > > disk controller. Disk controllers are less likely than > > motherboards to contain the latest technology. > > Um, and there is no reason why you can't just create a cabinet in which > the "main" processor *is* the "disk" processor.
Yep. We are back to a file server, eh? > > Once you toss out the RAID hardware you have both a simpler > > and a less expensive system. Nothing but a bunch of disks. > > This will almost certainly be more reliable than hardware > > RAID. ZFS provides _all_ the data integrity needed. > > More reliable? Not necessarily true. In reality, the fact the ZFS > checksums everything from start to finish buys you that irrespective of > the underlying implementation layer. Less reliable individually but cheaper hardware. But without the hardware relaibility of the RAID controller. JBOD. And as you note ZFS (software) is the source of overall system reliability. <quote> The job of any filesystem boils down to this: when asked to read a block, it should return the same data that was previously written to that block. If it can't do that -- because the disk is offline or the data has been damaged or tampered with -- it should detect this and return an error. Incredibly, most filesystems fail this test. They depend on the underlying hardware to detect and report errors. If a disk simply returns bad data, the average filesystem won't even detect it. </quote> http://blogs.sun.com/bonwick/entry/zfs_end_to_end_data This just sounds right to me. I think I am ready to drink the koolaid :) > The big advantage is that ZFS RAID trades some performance (write > barrier by readback and verify) for hardware. That puts the > "inexpensive" back in RAID again. > > -a Well I gotta say this appeals to the CB in me. BobLQ -- [email protected] http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
