Well yes Shane
I had a rant , and i guess i'll have more rants in the future ( i am not the
patient type when money is the subject :-)) )
I happen to be now in charge of the system department (all platforms) so i
have a HW ands SW budget for all platforms .
Although i am a techie and an MVS native at that , i have economics to take
care now and i have to decide where the money goes .
The talks we see every now and then on the list ( about the so called low
costs of mainframe VS the rest of the computing world ) shows that most of
the time the people ( not all) talking about it do not pay the bills
and do not run the servers themselves and have never seen a budget ) .
Mainframe simplifies my life and the life of my colleagues ,but it does not
satisfy my employer's wallet .Definitely.
Having said that ( take it easy please) , Big Blue discovered that we ( the
people deciding where we spend our money) were pulling out of mainframe
whenever we could . But then i think that is also what they want . You do
not keep such crazy policies without a goal !
It is not because of the platform or technical issues or complexity or lack
of competent people , etc ,etc .. ,but only for money reasons , pure plain
money reasons nothing else than money reasons .
( if Microsoft had followed the rules of MSU's billing , you'd have to pay
your windows XP home edition on your now 3.6 GHZ AMD 64 about one million
dollars :-)) ), but then you would have gone to Linux !
Today lots of companies are running their DB2 , their CICS their IMS
,because they can't get rid of thousand of cobol applications with a snap of
the fingers .)
The PVU is here to compensate the exodus and to size the opportunity offered
by ever greater and more mowerfull distributed platforms . ( 2 x dual proc
blades can easily replace an IFL )
The logic is that if you increase the power of your Intel or Sun machines (
hello Moore ) they charge you Moore ( does it sound like a joke in real
english ?) ,and you'll have the choice of paying more on mainframe or more
on Intel or Sun . I call that a lose-lose situation , and i guess they call
it a win-win one )
I requested from IBM and got a presentation about these¨Passport Advantage
PVU's last week because i needed to do some forecast on our 2007 budget
It is quite simple really : Before , on distributed systems , they would
charge you on the number of physical processors but now a server processor
is at least a dual core .Something had to be done .So they assigned a value
to a core , gave it a value that represents ( 50 for example) half of an
actual single processor ( 1=100) .They do this to keep the bill at its
present rate . ( they do not want us to jump to the roof right away or did
you say rant ? )
But they may change the value for one core in the future ( with the MHZ race
for example make it 70 ? ) .
By the way this billing ( before and after) does not care if the software
billed is mono processing single threading etc , it is just billed 4 times
on a 4 engine machine .
A bit of back to earth logic now :
Despite what some people ( IBMER's ? mainly ? ) say on this list on other
threads , the price of mainframe is not going down and this is why a similar
billing process is applied to distributed systems as it is a very profitable
scheme .
My total software and hardware mainframe bill ( like for everyone) is going
up each year despite the removal of a lot of applications .
Why is that so ?
Because the IBM theory of lower MSU's for more powerfull machines is a
pencil pusher theory instead of a production site reality .
Using bigger machines , they charge more on some product because they are
not always billing according to the consumption but according to the size of
the box on which it runs .Do not believe that you'll pay ( TSM was an
example) according to the size of the lpar where it is running on , you will
pay according to the total size of the box where you are running it , so do
not run it on a 32 engine box :-)) .
Your WLC bill changes when some mad programmer makes a loop on a 600 mips
engine instead of a 120 cmos box and the 4 hours average remembers perfectly
It is also increasing because a CICS v3 is more expensive than a V2 and
because a DB2 V8 needs more cycles than a V7 and it costs a lot more because
in order to benefit from the new MSUs of a Z9 , you'll have to spend few
million dollars to replace the Z/990 earlier in your planning .
Show me a shop whose mainframe total bill really went down and i'll show you
a shop that moved a part of its load somewhere else .
You realise when working on a 2 company merge that if you plan to double
your CPU size some of the software will cost you a lot more despite no
increased usage (not only IBM , ISV's as well) . ( a bit like paying 20
dollars the gallon of petrol because you changed the size of your car engine
, and having to bargain each time you go refuelling , crazy world ! )
IBM realised that things like Vmware with Vmotion are so spectacular that
standard VM or Lpar in mainframe could soon look a bit weak in comparison (
i talk about what it can do ) and that soon or later , we will move
more and more load to these platforms . ( i encourage you to observe the
Vmotion transfer of a production partition to another partition in another
machine in the next computer room without affecting the users )
( yes i mean a technology like TDMF or FDR/PAS brought to the processor and
storage part of your installation )
At first i did not believe it !But today i use it , and forget the urban
legends :these are servers and you don't need rebooting like you sometimes
do on your workstation . They became quite strong and solid .
So what IBM is doing is simply foreseeing the future , compensate the
departure of some mainframe shops , profit from Moore's law and the dual
and quad core trend . Simple .
I guess i'll get flak now , but then the list is here to exchange point of views
Bruno
Bruno(dot)sugliani(at)groupemornay(dot)asso(dot)fr
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
