Buck,
What a passioned reply! You raise a lot of issues - make a lot of valid points! Programmer productivity. Migrations from home-grown solutions to pre-packaged. Commodity pricing. The only point I might quibble with is continued growth of Microsoft. It appears to me that Microsoft never really figured out "cloud" computing, and is moving back to distributed server & desktop computing models.
The IT Director of the Washington Post recently asked for my opinion about the Amazon and Google models for cloud computing. I should admit that I really didn't understand them. I indicated that all our software runs on a server, and interfaces with Web browsers - but I imagined that a lot of organizations would be concerned about security and access of their data in the "cloud".
His reply kind of surprised me. He said frankly that he would NOT have those concerns, provided that they were addressed contractually. His only concerns would be performance, reliability, and price.
The Windows splash screen and musical jingle may be a nice marketing trick, but there is more substance to the idea of wirelessly connecting personal devices into cloud-based services, running like utilities, accessible from anywhere.
I personally relate more to Google's message for both clients and servers. With respect to servers, providers and users of services should not be concerned about hardware platforms, operating systems, application server configurations, database server configurations, performance, or scalability. Those things should just be taken for granted - accessed via simplified user and programming interfaces. And with respect to clients, Google is coming out with an OS that does little more than connect a browser to the cloud as fast as possible.
I'd like to see more hosting organizations providing cloud services under IBM i for both developers and end users, with the ability to integrate home-grown as well as pre-packaged software under a common runtime environment. And usage based pricing - like a utility.
-Nathan.
----- Original Message ----
From: Buck <kc2hiz@xxxxxxxxx>
To: midrange-l@xxxxxxxxxxxx
Sent: Fri, February 26, 2010 10:31:34 AM
Subject: Re: Moving forward or backward...
On 2/26/2010 10:30 AM, Aaron Bartell wrote:
I don't have numbers to support any of my theories, but I would be surprised
if your laptop could out perform in every workload. Not saying it isn't
possible, because I don't know - just that I would be surprised.
I wouldn't be surprised. IBM consistently sells atrociously
under-performing machines at a 'rock bottom' price. Ask anyone who
bought some of the original System/38s or AS/400s if they were happy
with the minimum configuration. I don't know of anyone.
Sure, Windows consistently under-reported their minimum requirements,
too, but Lukas' point is that the laptop he's quoting is far above the
Windows minimum requirements at a price that's impossible to match with
the IBM i minimum configuration.
That may be the secondary part of the message. The primary message is
that there is no standard workload emulator that you can run that will
tell you how a given platform will run compared to other platforms. His
'every workload' may not include a single one of yours, and all the
workloads the two of you run may not cover a single one of mine.
Seriously, how would any of us describe the workloads we run? Does a
word like 'batch' or 'interactive' really mean anything?
The endless debate of Microsoft vs IBM i is pretty much pointless
because the workloads aren't the same because the software isn't the
same. You can't run MAPICS or BPCS on Windows, you can't compile your
ILE RPG on Windows, you can't port your ILE RPG to Windows. Likewise, I
can't run Access or Excel, etc. on IBM i. Note that I'm talking about
SPECIFIC workloads. People don't simply switch from MAPICS to the
Windows equivalent, it's a gut-wrenching migration.
We use IBM i because we've always used an IBM midrange for our
workloads. We've always used it because our business was small when
System/3 was available, and IBM sold us a System/3 instead of a
System/360. We stayed with General Systems Division (aka midrange) ever
since because frankly, the conversion costs are too high to move off.
For newer workloads like email, we use Windows because IBM took forever
to bring TCP/IP to the midrange market space, and the native email was
not only late but it was and is way buggier than the cheapest freeware
email server for Windows.
This will probably get me some hate mail, but people need to wake up and
smell the lutefisk. We use BOTH IBM i and Windows. We won't be
converting from one to the other for any conceivable workload unless and
until the current software becomes so old and broken that it's cheaper
to replace en masse than it is to maintain. At that moment, the rarity
of IBM i expertise is going to be the deciding factor in what platform
gets the business.
If IBM wants to keep the batch and interactive workload business, they
need to keep me flush with cheap, easy to use tools so I can make my
boss ridiculously happy. If it takes longer and longer to maintain my
current software base, he's going to get less and less happy and it'll
always be tickling him that he *might* be able to buy a Windows package
lock stock and barrel for less than he pays me to keep the old one running.
If we want the platform to move forward, that's the leverage we'll need
to do make it happen. Pitting the platform against Windows is a losing
proposition for many reasons, not the least among them is that
executives use Windows every day and they never use green screen at all.
The i remains far in the background, but Microsoft gets their name in
eyeballs every time the PC plays that blasted 'I'm booting' jingle.
I want to finish up with the note that I have uttered many a blue word
at Windows, Windows programs, Windows programmers, the amateurish
hardware that Windows runs on and the whole Windows fan boy crowd in
general. I'm not interested in migrating to Windows at all. I very
much like my IBM i and I fully plan on programming in RPG for the next
20 years - that'll make 50 for me.
So I'm not a Windows cultie, but the simple reality is that as we RPG
programmers get older and scarcer, Microsoft is going to gain market
share for several reasons:
1) Someone wrote a Windows package that 'does that.'
2) There aren't any RPG programmers left.
3) The IBM i packages do less.
It's been an interesting couple of threads.
--buck
midrange.com
