On 12/23/09 10:11 AM, "Mike Walter" <mike.wal...@hewitt.com> wrote:
> In SFS, the SFS server manages a catalog mdisk. That catalog has a pointer to > every 4K block in the filepool. When a (large or small) is erases, the block > pointer for every file needs to be cleared, and IIRC, a bit in every block > that the made up the block. Without looking at the PTF, perhaps something in > that process was updated to speed things up, just guessing. Don't forget also that SFS has to manage a APPC connection and a dispatch of another virtual machine involved. Much longer code path for similar operations.