mitchd...@gmail.com (Dana Mitchell) writes: > Just like buying a z CPU and only paying for the number (and capacity) > of CPU's actually turned on via microcode. Things never change!
re: http://www.garlic.com/~lynn/2014m.html#87 Death of spnning disk? http://www.garlic.com/~lynn/2014m.html#88 Death of spnning disk? http://www.garlic.com/~lynn/2014m.html#101 Page Data Set Sizes and Volume Types the semi-facetious scenario at SHARE was charging more for smaller 3380 capacity ... the issue was fully loading 3380s could degrade overall system throughput, enormously more than offseting the savings in number of 3380s. the issue was datacenter managers weren't making the connection between better overall system throughput and partially loaded 3380s. the objective was to come out with a "different" 3380 model that solely focused on the improved overall system throughput (and charged more for it) as a way of resetting datacenter managers' point-of-view. since then, various kinds of staging and caching at many levels in the system has gone a long ways to obfuscating physical arm access latency (goes along with little dustup I caused in the early 80s with observation that the relative system throughput of disks had declined by a factor of ten times over the previous 15yr period, aka processor throughput went up 40-50 times, disk throughput went up 3-5 times). -- virtualization experience starting Jan1968, online at home since Mar1970 ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
