I needed a break from DIY, and my typing went a bit OTT here.
If anyone at RMplc wants any more information or guidance
then please email me directly - advice is free, and I'm not sure if this is OT.
On Fri, 28 Jul 2000, Simon Rainey wrote:
OK, you've got a developer base of several
hundred programmers who eat,
drink and sleep Microsoft class libraries, Visual C++, Visual Basic and
Developer Studio. They know NT inside out but can't tell sed from awk and
think vi is probably a character from Deep Space Nine.
A software developer is a software developer is a software developer. They
solve problems, and our skills are easily transferred across a whole range of
languages and development environments. I get very insulted if told that I
'know' a particular environment and therefore do not 'know' another
solve problems on it. Probably several hundred of your several hundred
programmers went to a University or equivalent and have experience of *nix
systems. They probably do not know NT and developer studio inside out, for 2
reasons. Firstly, it is over complicated, and secondly it is constantly
changing and being re-invented. MCPs have to constantly sit exams if they want
to stay certified, and software needs constant changing just to keep it the
same - how many times has there been the necessity for software to be updated
for a 'transparent' OS or developer tools update (which you are forced to do
just for a minor bugfix)? And you don't need to use sed awk or vi to produce
*nix applications - the command line just improves productivity for those with
the ability to use it.
On top of that the
Microsoft marketing behemoth is going into overdrive to promote Win2k *and*
to belittle Linux along the way.
And for once it is being ignored. Huge amounts of money was spent
on replacing PC kit to make it Y2K complient, which came with the latest
versions of MS software which (arguably) is fit for purpose. Very little money
is being spent on OSes this year. The next major spend will be in 2003, by
which time there should be a choice between an expensive OS which requires high
spec hardware, and another OS we all know and love. If not by then, then it
will be in 2006.....
With Microsoft intent on polarising the
market you need to back Win2k or Linux. You can't do both - it's too
Sounds like you need to replace your advisors here - you can do both, and it
saves money. The only people saying otherwise are those who stand to lose
Take a good look at the MS roadmap. The next OS to expect is going
to be based on wonderful new technology that the world has never seen before.
Hmm. Actually, the next OS is going to be a thin-client, as MS have realised
the threat from server-side applications. Have another look at the roadmap.
What is missing? If you haven't spotted it yet, then have another look at what
is new. C#, which is pronounced 'C hash' by computer literate people, and
sharp' by musicians. You will be expected to retrain your developers to use
C#, and that is going to cost money. What benefits will that give you?
Apparently it is cross platform, although it will only ever succeed on the MS
platform, as MS will probably change it so often that other vendors will give
up. By now you should have noticed what is missing. Java. MS couldn't buy
it, they couldn't hijack it, so now they are pretending that it doesn't exist.
It does exist, and it does everything it says on the tin. You can even
download development tools for free to give it a try. (Have a look at Visual
age for Java from IBM). You have all the benefits of C#, but you are now
backing both horses, and with current hardware Java performance is not the
issue it once was.
and besides, your established user base won't
apparent change of direction. You've been selling them NT for years and now
you want them to change to Linux?
The problem here is that they will want to know why you have been selling them
NT for years. In fact, this doesn't have to be a problem. All the user ever
sees is the interface to the program. They don't mind what it is written in,
or if the back-end is Oracle, DB2, or SQL server, hosted on NT, Netware, or
Linux. If you move some of the funtionality to the server, then what would they
care? Only that it would save them money as their desktop has a longer useful
life, and the server resources are a more cost-effective upgrade. Start that
process now, learn about Java and Enterprise java beans - you've got nothing to
lose. Otherwise, make sure that your applications run under WINE as an
And all the while your investors are
keeping a close eye on your performance against predictions made up to a
year ago. Overspend without a convincing story and your share price will
plummet. It's a tough call.
Hmm. You could actually be going out of business. IBM refer to Linux as a
'disruptive technology'. That means that it is such a big change, and so
important, that they are prepared to re-engineer their entire business around
it. What are you doing? Remember, for a zero cost you can back both horses.
All I can say is that RM does listen. If enough users
make serious noises
about Linux to their account managers then things may well change.
Otherwise NT is here to stay.
The earth is flat, cars will never catch on, and if we were meant to fly we
would be born with wings. NT only has a limited shelf life, as does W2K. We are
not that far away from any OS being able to run any application - who would
want NT then? Microsoft are aware of this, as are all other OS vendors.
Microsoft will be around forever in one form or another, but now that they are
being forced to separate their applications from the OS, the programming
interfaces can be very easily translated to native OS calls on any platform.
The WINE project is now aiming at a relatively fixed target, for example.
If RM only follow then other vendors will start to lead. Here is a news
announcement which should give you the roadmap to the real future of
computing IMHO, and I would suggest that you do some serious cost-benefit
analysis, and get your IT director nervous - he/she should be, as his/her
future is at stake.
For info., SuSE do have a Linux distribution for the computer mentioned in the
following (currently in beta). Also note that the machine in question is
relatively low-spec. How about one machine to serve every pupil in the
country? One day it will be possible, and the economies of scale don't get any
better than that. When reading it, bear in mind that Star Office is now
available under the GPL, and we are talking about application servers, not just
http and email. Note also that the figure of 41500 may not be a real limit -
the machine was not optimally configured. $125,000 is also roughly what it
costs to employ a developer for a year (not what they are paid, I might add!),
so the figure is not as big as it may seem.
The open source IT Express is
sponsored by SupportSource.com
the open source IT Express! July 31, 2000
The open source IT Express is a service of EarthWeb Inc., a
public company. For information about our company, please
go to http://www.ewbx.com
The $45 Linux PC
This week, IBM will announce a new pricing scheme for Linux
on the S/390. By the end of September, you will be able to
buy a mainframe engine that runs Linux for $125,000. That's
a third of what it costs for one that runs OS/390.
For $20,000 more, you can buy software that will let you run
multiple copies of Linux on the same machine. IBM is calling
it S/390 Virtual Image Facility or VIF. (You can do the same
thing under OS/390 and VM, but that costs more).
How many instances of Linux can you run on one machine?
David Boyes, a consultant who works with the S/390, managed
to boot 41,500 Linux servers on one mainframe. You may not
be able to run that many in real life. But Boye's company,
Dimension Enterprises, in Herndon, VA, has one telecom
customer which has 3,200 copies of Linux running on one S/390.
At $145,000 for the mainframe engine and license, that works
out to $45.31 per (virtual) Linux box.
Try buying a PC for that.