IBM Z and LinuxONE IBM Z

IBM Z

The enterprise platform for mission-critical applications brings next-level data privacy, security, and resiliency to your hybrid multicloud.

 View Only
  • 1.  Comparison IBM Power processor to IBM z LinuxONE IFL

    Posted 5 days ago

    Hello,

    Is there any comparison of IBM Power processor to IBM z LinuxONE IFL ?

    Any type of comparison is ok, benchmark, just raw strength etc.

    Thank you,



    ------------------------------
    Arye Shemer
    IBM z
    Tangram-Soft (Matrix)
    ------------------------------


  • 2.  RE: Comparison IBM Power processor to IBM z LinuxONE IFL

    Posted 5 days ago

    What is it that you want to compare?

    Z is a CISC processor, POWER is a RISC processor - they come at solving problems in a significantly different way.  Ultimately your needs for bigger/better/faster are going to come down the application(s) you (want to) run and how well they are implemented for the platform and how that platform is realized as a "computer"...then things like "how good is the compiler for <implementation language>" and "how is I/O handled" (along with a whole host of other architectural concerns) become the next questions.

    Looking at gee whiz numbers at the processor level is rarely useful, similar to looking at raw engine numbers without the rest of the car.



    ------------------------------
    Scott Fagen
    Mainframe Evangelist
    CDW
    www.cdw.com/content/cdw/en/solutions/ibm-zsystems.html
    ------------------------------



  • 3.  RE: Comparison IBM Power processor to IBM z LinuxONE IFL

    Posted 4 days ago

    Thank you Scott,

    The question came from a customer which express his interest in comparison of Oracle database running on Power against Oracle database on LinuxONE.
    We would like to offer the customer POC on current IBM z  customer machine,
    but before that we would like have some strength comparison in order to understand if we have enough unused engines (to make them IFLs)  for the POC.
    I hope that make my previous post more sensible.

     

    Thanks again and best regards,

    Arye Shemer.

     

     






  • 4.  RE: Comparison IBM Power processor to IBM z LinuxONE IFL

    Posted 4 days ago

    Hello, I am also interested in this benchmark. We started a POC that aims to run a Java application via OPEN Shift on a Linux One. But we already have a lot of Power for DB2 LUW and a little OPEN Shift and the question comes up often.



    ------------------------------
    François DELTOMBE System engineer
    System engineer
    CA-GIP
    Gradignan
    0623788908
    ------------------------------



  • 5.  RE: Comparison IBM Power processor to IBM z LinuxONE IFL

    Posted 3 days ago

    Hi, 

    Scott entry is fully right. My two cents here is that you will not find that kind of benchmark because there's no sense in running it. The fact that both systems can run the same operative system, no matter the flavour, doesn't mean that you can compare both systems. They are architecturally different and built for different purposes. 

    Designing, manufacturing and supporting two different platforms is a big effort for a company. If both platforms would be aimed for the same tier and for the same workloads, it would be a nonsense to keep them both in the catalog. 

    IBM Z architectural choices are intended to work with data, to support I/O operations as fast as possible without interrupting the execution of processes that are intended for business. As an example, in the very last z17, I/O has been brought into the same chip where cores are working, to be able to fill the huge L4 to L1 caches. It's called Data Processing Unit (DPU). The architecture of IBM Z is designed to multitask and handle different processes doing different things at the same time. Power architecture is designed to run processes in the same core for as long as possible until it is finished.

    CISC processors are equipped with microcode and specific instructions for some operations, RISC processors are intended to have a smaller set of fixed instructions linked to the physical wires of the chips.

    I just wanted to name a couple if things, there are others, that justify that both platforms are incredibly good for what they do and there is no point in comparing them. You can't use a cruise ship to move containers (physical ones: big metal boxes with things inside) and you can't use a container ship to move passengers. They are both ships but intended for different tasks. 

    Regards, 

    D.



    ------------------------------
    Diego Cardalliaguet
    IBM Data&AI zStack Leader for EMEA
    IBM
    ------------------------------



  • 6.  RE: Comparison IBM Power processor to IBM z LinuxONE IFL

    Posted 3 days ago
    Edited by Steve Able 3 days ago

    Benchmark white papers are good, but like others have pointed out in this thread - the raw numbers are like comparing "Apples to Oranges"... 

    They both are fruit - but until you actually try - you will not know which is best for you!

    Am partial to IBM Z myself and IBM is pushing the LinuxOne Big Time!  So, you might be able to find an IBM sponsor that can provide a side-by-side environment in order to run actual benchmarks of the environment you need to prove or dis-prove.

    A few years ago, an IBM Sponsor got an environment setup for a Service Enablement test of performance with CICS/TS and IBM z for our company. This allowed us to gain some very useful insight to how well CICS and z/OS performed. 

    The other issue you have to consider which most people overlook... Does the target client have the skills or are they willing to pay someone to setup and maintain their in-house environment.  Does the client have Linux z (LinuxOne) skills or are the majority of the skills on the Power?

    People cost to maintain, and performance tune the environment is often overlooked. During our Service Enablement Performance test, we discovered several configuration changes that greatly improved the thruput. 

    Tweaking the performance settings on each platform most often will change the results. 

    But with any Performance Test - your mileage might vary - our performance runs did not have anything else executing in the environment.  

    Good Luck!

    ☠ 💀 ☠ 
     



    ------------------------------
    Steve Able
    Director of Strategy and Architecture
    Adaptigent formally GT Software, Inc.

    ------------------------------



  • 7.  RE: Comparison IBM Power processor to IBM z LinuxONE IFL

    Posted 6 hours ago

    Great conversation to be a part of.
    There is RISC vs CISC as part of the conversation.
    There is proximity of the data to the CPU ( Both Z and Power make great strides in this through every iteration of chip )
    While CPU speed is faster in the LinuxONE  (5.5 GHz ) vs Power11 ( 4.4 GHz) , do not forget about SMT ( Symmetrical Multi-Threading ).  LinuxONE is still operating on SMT-2 ( 2 threads per core ) whereas POWER has been SMT-8 ( 8 threads per core) since POWER8 in 2014.
    LinuxONE has made great strides in ease of use with DPM ( Dynamic Partition Manager ) for provisioning which has a VERY similar look and feel to provisioning from a POWER HMC, meaning that your POWER skillset can more easily translate into usable skills in a LinuxONE environment. 
     



    ------------------------------
    Alan Fulton
    Follow me on Twitter - @The_Iron_Monger
    ------------------------------



  • 8.  RE: Comparison IBM Power processor to IBM z LinuxONE IFL

    Posted 5 hours ago

    Thank you all for contributing to this discussion.

    It help us a lot understanding the issue of comparing the of those excellent machines.

     

    Thank you and best regards,

     

    Arye Shemer