On Mon, 14 Jul 2003, Lars Holmstr�m wrote: > I am not entirely familiar with Linux clustered filesystem. Maybe some one > could point me in a direction for a real implementation for a disaster > tolerant mailserver ? > > As far as I understand the inode modification concept require applications > to be linked in to new kernel code ? > > Some other file systems (like GFS) requires specific hardware as far as I > understand (iSCSI or FC) and most of the Clustered file systems simply share > the storage, usually SCSI and would sometimes be somewhat limited.
iSCSI has nothing to do with hardware. iSCSI is "Internet SCSI". It's SCSI over Internet. On Linux you can easily use 'hdc=ide-scsi' or some such, but I doubt that's necessary. With iSCSI, hardware doesn't really enter into the equation, anyway -- it's just providing a block device over the internet, using a variation of the SCSI protocol. The source of the block device shouldn't matter. FC is a hardware thing, but honestly, I can't see what it might have to do with GFS. Storage is storage. IDE to FC stuff is cheap these days. > The benefit with my concept would be that there are no kernel modifications > and that the synchronisation may run over existing networks, and even slower > lines that 10 Mbps, depending on application load. -- No more monkey ducks, jumping on the bed! Jon Nelson <[EMAIL PROTECTED]> C and Python Code Gardener ------------------------------------------------------- This SF.Net email sponsored by: Parasoft Error proof Web apps, automate testing & more. Download & eval WebKing and get a free book. www.parasoft.com/bulletproofapps1 _______________________________________________ courier-users mailing list [EMAIL PROTECTED] Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users
