We still have the concept of a capture ratio - except I expect it to be in the region of 85 - 95%. If it's not then something's up. (It's one of the things I chart.)
Cheers, Martin Martin Packer, zChampion, Principal Systems Investigator, Worldwide Banking Center of Excellence, IBM +44-7802-245-584 email: martin_pac...@uk.ibm.com Twitter / Facebook IDs: MartinPacker Blog: https://www.ibm.com/developerworks/mydeveloperworks/blogs/MartinPacker From: Anne & Lynn Wheeler <l...@garlic.com> To: IBM-MAIN@listserv.ua.edu, Date: 06/15/2012 03:38 PM Subject: Re: How many cost a cpu second? Sent by: IBM Mainframe Discussion List <IBM-MAIN@listserv.ua.edu> shmuel+...@patriot.net (Shmuel Metz , Seymour J.) writes: > It may be the only server for which the OS provides accounting > information, but that's not the same think. We don't know how much of > the cost of the box is in support of instruction execution, how much > in support of memory and how much in support of I/O. Then there is the > issue of allocating DASD expenses. Ditto environmentals and staff. > > What makes the mainframe marginally easier is that there is a smaller > number of boxes and they all come out of the same budget. re: http://www.garlic.com/~lynn/2012h.html#59 How many cost a cpu second? http://www.garlic.com/~lynn/2012h.html#60 How many cost a cpu second? MVS also used to have capture ratio ... that it actually could only account for 40-60% of use. There would be a process where all the unaccounted for cpu time (total cpu found by taking elapsed time minus wait time) would be prorated across the accounted for use. This is similar to all the other total cost of ownership for running a datacenter ... getting prorated across the accounted for use. Long ago and far away ... we ran into it looking at moving some large applications from mvs/3033s to large number of vm/4341s and trying to estimate how big the increase in throughput would be. part of the issue was that the mvs/3033s were running approx. 50% capture ratio (and had initially been unaccounted for in the considerations). example of charging ... this references Amazon cloud "on-demand" charging $2.40/hr for "CC2" (although better bulk rates are available, CC2 is e5-2690, rated 577BIPS about equivalent of eleven 80-processor z196 rated at 50BIPS) http://www.pcworld.com/article/251536/amazon_hails_era_of_utility_supercomputing.html this references somebody doing on-demand cloud, supercomputer at 240TIPS (240,000BIPS, about equivalent of 4800 fully configured 80 processor z196) http://news.cnet.com/8301-13846_3-57349321-62/amazon-takes-supercomputing-to-the-cloud/ part of "on-demand" service is that they have to over provision for peak demand ... might have 50% idle at low-points. It figures into total cost of ownership that has to be prorated across charged for service (in manner similar to MVS capture ratio). However, there can be special "off-peak" rates for using resources that would otherwise be idle. -- 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 Unless stated otherwise above: IBM United Kingdom Limited - Registered in England and Wales with number 741598. Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN