One calculation from real life I have seen is that the z alternative hw cost was just below that of Intel blades per server. And also that hw cost is less than 10% of total cost, for all platforms.
Just be sure to have ALL costs included. It's sometimes hard to get that in the Intelx86 case. Cordialement / Vriendelijke Groeten / Best Regards / Med Vänliga Hälsningar Tore Agblad Volvo Information Technology Infrastructure Mainframe Design & Development SE-405 08, Gothenburg Sweden E-mail: [email protected] http://www.volvo.com/volvoit/global/en-gb/ ________________________________________ From: Linux on 390 Port [[email protected]] On Behalf Of Rodger Donaldson [[email protected]] Sent: Thursday, June 18, 2009 11:29 To: [email protected] Subject: Re: FW: Rules of Thumb Harder, Pieter wrote: >> I agree that the speed of CPU is not the problem anymore. But do you want >> CPU intensive load? Afaik the CPU in >mainframe, even z10, is still more >> expensive than an intel. So if you'd have CPU intensive loads, would it be >> still cost >effective to run on z? Do we still need to stay away from these >> applications? > > Just talking of the top of my head (meaning no real life comparison > experience, yet..) > > When you run out of the capabilities of standard Intellish hardware, you have > three options: > - splitting a two-tier setup to three-tier. Expensive/more complex because of > multiple server images > - look at non-standard Intellish hardware that is probably a lot more > expensive > - or look at something completely different like z or p (will Sparc still be > there to look at?) > > Of course it all depends, but option 3 may not be so far out as you think at > first glance. To amplify: Think about what if actually costs to deploy on decent hardware. My standard Intel server would require two power supplies, two single port FBA cards (if SAN attach is a requirement), and four ethernet ports (2 x failover primary, 1 backup, 1 management). Once you start eating a couple of power points, 4 switch ports, and two fibre ports per server, the cost, if accurately accounted for, looks quite different. Then, of course, you start wanting to cluster for availability, and the cost looks even less attractive, especially since you will be buying enough hardware that you can take full load on n-1 nodes. ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
