In theory, MSUs are the right answer, because they're supposed to be normalized reflections of real-world throughput. OTOH there's the "technology dividend", where a new machine's MSUs may be "larger" than the previous machine's--IOW, n MSUs on a newer machine may net you better performance than the same n MSUs on an older machine. (Or maybe this is just because the old machine is getting rusty :) )
And of course there's the fact that different workloads perform differently. But MIPS are even fuzzier and are thus a worse measure. In theory, there are IBM sales reps who should be able to help with this, too. -----Original Message----- From: IBM Mainframe Discussion List <[email protected]> On Behalf Of Jay Maynard Sent: Friday, June 12, 2026 10:51 AM To: [email protected] Subject: Re: Selecting z model MSU When we did this last year, we simply picked the capacity level that was the closest match to the z14's and wasn't any smaller. We wound up going from a z14-L02 to a z16-M02. It's worked out well for us. On Fri, Jun 12, 2026 at 8:27 AM Jake Anderson < [email protected]> wrote: > Hello > > Are there any specific headroom for MSU to be considered while moving > from > z14 to z16 > > For example , > If we have 55MSU, What would be the formula or a best approach to > select a > z16 ? > > Considering based on MIPS is a good idea or an MSU ? > > Any pointers would be appreciated > > Regards > Jake > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, send > email to [email protected] with the message: INFO IBM-MAIN > -- Jay Maynard ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
