On Mon, 10 Jun 2013 11:38:08 -0400, Farley, Peter x23353 <peter.far...@broadridge.com> wrote:
><Rant> >Like a few others on this list, I have often gritted my teeth at the necessity >to estimate disk storage quantities that vary widely over time in a fixed >manner (i.e., SPACE in JCL) when the true need is just to match output volume >to input volume each day. > >Why is it that IBM (and organizations that use their mainframe systems) so >vigorously resist a conversion off of the ECKD "standard"? (Yes, I know it's >all about "conversion cost", but in the larger picture that is a red herring.) > Not that I'm likely to see such a transition in my lifetime, but in this >dawning time of soi-disant "big data", perhaps it is past time to change the >storage paradigm entirely, not from ECKD to FBA but to transition instead to >something like the Multics model where every object in the system (whether in >memory or on external storage, whether data or program) has an address, and >all addresses are unique. Let the storage subsystem decide how to optimally >position and aggregate the various parts of objects, and how to organize them >for best performance. Such decisions should not require human guesstimate >input to be optimal, or nearly so. Characteristics of application access are >far more critical specifications than mere size. The ability to specify just >the desired application access characteristics (random, sequential, growing, >shrinking, response-time-critical, etc.) should be necessary and sufficient. > >EAV or not EAV, guaranteed space or not, candidate volumes, striped or not >striped, compressed or not compressed - all of that baggage is clearly >non-optimal for getting the job done in a timely manner. Why should >allocating a simple sequential file require a team of "Storage Administration" >experts to accomplish effectively? ></Rant> > >Peter Oh, then you want to move to IBM System i... :-)> Seriously, System i, (formerly known by many different names), addresses everything in "storage/memory" and on disk and has been 64-bit since the mid 1990s, (and it was 48-bit before that). There is however, no need for system programmers on the i, (really IBM, you can't come up with better names for hardware/software than 1 character? And "i", is this an Apple box?) -- Dale R. Smith ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
