Interesting thread. Everything i've measured says a megahertz is a megahertz. MIPS are not always MIPS though at least for cross platform comparison. I put the IFLs in a range of .8Ghz to 1.2Ghz. There is an approximate MIPS/Mhz comparison - i've seen this published on an IBM foil as "barton's number": 1 MIP is close to 4 Mhz. This is free, i didn't pay anyone for it. validate for yourself. I have yet to see anyone try and disprove this with real numbers, lots of hand waving going on tho.
The only processor that i don't trust this is the 990 that has some performance enhancements for Linux i don't understand. Maybe the number's 6 or 7, it's not 10. For I/O, I think most Unix platforms will give a "z" competition for ONE application. If you have Escon channels, don't even try and compete with "z". Ficon express channels are very competitive. The amount of CPU used for I/O is FUD - for ONE application. The amount of concurrent I/O managed on "z" is something else. >From real data i've looked at, many servers out there run at 3-5% utilization. But most installations don't bother measuring to validate - why measure CPU utilization on a system that during prime time averages 3-5%? who cares? Why 3-5%? One does not run muliple applications under WinNT, not even a print server and a file server at the same time because the multitasking is poor, one application will take over and other applications are slow. If this has changed give my apologies. Oracle is always interesting. Some databases i've been looking at recently average something less than 1% of CPU. If you have 30 servers on intel running oracle, moving the lower 25 of them to Linux on zVM saves a lot of money in licenses and may not even drive one IFL. Move the top 5 and you may need 10 IFLs to support them - maybe not a great move. Too many consolidation projects are failures for not recognizing Mhz is Mhz. If your application peaks at 25% of an 800Mhz Unix server, moving to Linux on "z" is a no brainer. Moving a workload that will consume 30 Dual 3Mhz processors just is not viable. Having a successful consolidation is not rocket science. Measure the servers for a long period of time, understand the peaks, understand the averages. Multiply the peak Mhz requirement by 4, that is your MIP requirement. Move some servers, validate the plan, move the rest. Why does "z" make sense? driving "z" processors to 80% makes sense - and still meets service requirements. Other platforms run applications that are not used all the time, but they only run one app per server so lots of unused cycles. The "z" uses the cycles. I could go on. If anyone wants to have a successful consolidation by the book, and show the numbers, send me a note. Would be nice to publish the numbers for at least one that didn't leave gaps leading people to believe that a 4-way 1.2Ghz server could easily be replaced by one IFL. Only if it peaked at 20% utlization.... >Date: Thu, 30 Jun 2005 15:18:22 -0400 >From: David Kreuter <[EMAIL PROTECTED]> > >Many good points raised by all. Lots of benchmarks fail because >customers are led to comparing non-z workload with "plunk and go" linux >vrtual machines. Do this only if you want to make z series look bad. >I've seen lots of numbers, and done some benchmarking. Z series >typically does not win in a CPU to CPU comparison versus price point. >There are other significant wins for z series, many of them mentioned in >this thread. As far as I am concerned there is a significant processor >cost for linux work on z series. Discuss deployment, networking, >personnel resources, TCO, complexiity,even I/O - but CPU? - thin ice. > > I have rarely seen CP overhead be an issue. I find the 10% for CP >overhead way high - maybe 3% - 5%. >David > >>Joseph Temple wrote: >> >>Part of the difference is the comparison of the HP to the Intel >>Box. If you take the z900 uniprocessor IFL to be 256 MIPS my >>first guess for = the HP 440 MHz 2way would have been 290 MIPS, >>but for the PIII 700 MHz I wou= ld have come up with 83.5 MIPS. >>I do these comparisons by using extrapolati= on data we purchase >>from Ideas International or using Internal IBM extrapolations >>(they yield similar results but cover different machines)= , and >>then apply a middle of the road conversion factor and a 70% >>utlizati= on factor. Any way the 290 MIPS would chew up an "IFL >>and then some". I can't share the conversion factors with you >>because of agreements IBM ha= s with benchmark councils and with >>Ideas. In any case I think your estimator is over valuing the >>PIII 700 MHz engine. >> >> >>Joe Temple >>Executive Architect >>Sr. Certified IT Specialist >>[EMAIL PROTECTED] >>845-435-6301 295/6301 cell 914-706-5211 >>Home office 845-338-1448 Home 845-338-8794 >> >> >> >> >> >> >>I would guess that the SQL is not very optimized. Also, the app >>pushes = a lot of math that could be done client side to the >>database. >> >>700 may be a somewhat low, but the HPUX server had two 440mhz >>PA-RISC 8?= 00 cpu's. When we moved it, it ate all of one IFL, >>plus a bit. It's bloat= ed since then and has been chewing on a >>good portion of both recently. >> >>-----Original Message----- >>From: Rich Smrcina [mailto:[EMAIL PROTECTED] >>Sent: Thursday, June 30, 2005 12:15 PM >>To: [email protected] >>Subject: Re: Business Week Article >> >>700Mhz? That seems pretty low. >> >>Little, Chris wrote: >> >> >>>Were those Solaris/AIX servers underutilized? By a lot? >>> >>>Certainly workloads don't compare across platforms. I agree with >>>that, but for database workloads I was told (informally) to assume a >>>single z/900 IFL at a Pentium III 700mhz. Within reason, it seems >>>correct. We are probably getting better than that, but certainly nowhe= >>re >>> >>> >>near what you are seeing. >> >> >>>Low CPU/high IO (webserving, maybe?) might translate better to linux >>>on zseries, I don't know. >>> >>>-----Original Message----- >>>From: Uriel Carrasquilla [mailto:[EMAIL PROTECTED] >>>Sent: Thursday, June 30, 2005 12:12 PM >>>To: [email protected] >>>Subject: Re: Business Week Article >>> >>>Crhis: >>>I suspect not all workloads are the same. What we are finding >>>out is that loads are taking place much faster. That was a large >>>pecentage of our usage before. We don't have any users signed on >>>and we only run the zLinux image to service requests. It works >>>for us and no one is complaining. >>> >>> >>>John: >>>The formula we use is 4xCPU (sun 1.2 GHz or aix 1.0 GHz) is the same >>>as one IFL under z890 running zLinux under LPAR. >>>If zVM, we take away 10% power for the zVM overhead. >>>This is a rough estimate and gets refined upon the real circumstances. >>>Products licensing based on number of CPU's get penalized but we spend >>>more for the H/W (IBM is happy). >>>There are other benefits that are more important, such as saving in >>>head counts that I rather not get into but you can figure it out. >>> >>>Regards, >>> >>>[EMAIL PROTECTED] >>>NCCI >>>Boca Raton, Florida >>>561.893.2415 >>>greetings / avec mes meilleures salutations / Cordialmente mit >>>freundlichen Gr|_en / Med vdnlig hdlsning >>> "If you can't measure it, I'm Just NOT interested!"(tm) /************************************************************/ Barton Robinson - CBW Internet: [EMAIL PROTECTED] Velocity Software, Inc Mailing Address: 196-D Castro Street P.O. Box 390640 Mountain View, CA 94041 Mountain View, CA 94039-0640 VM Performance Hotline: 650-964-8867 Fax: 650-964-9012 Web Page: WWW.VELOCITY-SOFTWARE.COM /************************************************************/ ---------------------------------------------------------------------- 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
