On Wed, 2 Jul 2008 17:49:06 -0500, Rick Fochtman <[EMAIL PROTECTED]> wrote: >------------------------------<snip>--------------------------- >How difficult would it be to write (or under the covers, convert) CKD >CCWs to FBA CCWs? >----------------------------------<unsnip>------------------------ >Isn't that exactly what happens in the current crop of RAID controllers now?
No. Count-Key-Data describes a *semantic* for accessing data. Many of those semantics (such as searching) do not apply to FBA. Another example: There is no RESERVE or RELEASE on in the FBA architecture. It is trivial to convert CCHHR to an FBA block number. z/VM, z/VSE, and Linux are able to take advantage of SCSI devices because their file systems treat the disk as a block device. You won't find physical FBA block numbers or CCHHRs in the filesystem interface. File system block numbers are converted to CCHHR or physical FBA block numbers, or memory pages, as needed. But the limitations of the FBA architecture mean that, for example, I cannot share a RACF/VM database when deployed on FBA. Even if CP's simulation of FBA on SCSI was updated to include RESERVE/RELEASE, it wouldn't work because there is no concept of channel path groups and no "lockout" mechanism in the SCSI controllers. Alan Altmark z/VM Development IBM ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
