It should be rather obvious to you, or anyone else at this point, that  
I was one of the people who benefited
from the free academic license offered for BSD in the early 1980s -  
and moreover, took advantage of it.

It is also just as obvious, at least to me, you have little or no idea  
of what you are talking about, have little interest
in holding a real conversation, and for some reason, known perhaps  
only to you, you have decided to target
me with your derision.

Just for the public record, I did mention to Mr. Eggers that I didn't  
bother reading his previous post and that I went
back and read it before replying to him in private. And I once again  
offered to take the time to show him what I was
talking about.

Mr. Eggers seems to feel he has some points, however, even a modest  
examination of his data reveals quickly
that he is making unwarranted conclusions based upon inadequate and/or  
false and misleading data, and basing
his opinion on what amounts to a poorly executed analysis of that same  
false and misleading data.

For the record: first, while Linux can and is used to do some  
fantastic things, it is able to do so at reasonable speeds
not because it is a very efficient operating system, but because the  
platforms it operates upon are very fast, and mask that
inefficiency. As workloads on Linux grow, this becomes very evident  
indeed.

Second, virtualization technology can and does alleviate some of this  
inefficiency, allowing systems like mainframes to run  hundreds of  
copies of Linux and use them for real workloads. I referenced  
Nationwide Insurance, which runs about 600 instances of Linux, with  
about 2500 JVMs running on those instances, on two Z10 mainframes.  
Doing real work - running Websphere, DB2, MQ and other products to  
support numerous applications. Virtualization is provided by a real  
native OS (z/VM) running what is perhaps the worlds best  
virtualization technology.

Third, I contend that operating systems written to take advantage of a  
native platform will always be more efficient, and at least in theory,  
able to handle more workload than an operating system written with a  
prime goal of portability. This is certainly true on the mainframe  
platform, and in the case of Windows vs. Linux, and I believe it to be  
true in the case of operating systems like MPE/ix. I believe it so  
with MPE/ix, but I don't have the chops to say much of anything  
authoritatively about MPE/ix.

What Linux offers is zero license cost and a very open platform with  
fantastic network capabilities.

Lastly, for whatever reason, Mr. Eggers seems to enjoy taking some  
personal shots at me. I am not sure why, but his rather nasty  
accusations of incompetence rather annoy me, so I will take two  
actions. First, anyone who would like to take the time to visit with  
and examine my environment is welcome to come down to Austin and spend  
a day with me. I'll be glad to show you some of the stuff I do, and  
more, show you easy to understand evidence of the claims I made  
above.  Second, I simply won't participate in this conversation with  
Mr. Eggers any longer, as it is a waste of bandwidth and probably not  
at all enjoyable for anyone.

I would greatly enjoy a discussion of these subjects however, so if  
anyone else would care to share their thoughts, I will read them with  
with great interest. I may disagree mind you, but I will read them and  
think about them.

Oh, and if anyone has been offended by this nonsense, please accept my  
apologies. I should have recognized a troll like this for what it was  
immediately.

-Paul

On Mar 6, 2009, at 2:22 AM, Peter M. Eggers wrote:

> Paul -
>
> I got your private email, and since it would be unethical to post  
> private
> correspondence on a public list, I will save you the embarrassment,  
> though I
> think it would be worth a few good laughs!
>
> Stating that you you only read the first few lines and then claim  
> that I
> hadn't debunked anything you said, is kind of silly, don't you  
> think?  The
> fact that you picked my comment about a Dell box being able to bring  
> up many
> virtual Linux systems which was in the middle of my post, tells me  
> that you
> actually read a substantial part of my post, if not all of it.   
> Getting off
> the topic of Linux's efficiency here, but does speak to your  
> credibility.
>
> Speaking of credibility, you are a Linux/Unix expert because you  
> made some
> sort of contribution to BSD in 1982?  LOL!  That was when BSD was  
> first
> getting networking implemented and the whole OS had several major
> overhauls.  Telling me that you will send me a 9-track tape so I can  
> "fgrep"
> your name?  Sheesh, just burn a CD and save yourself postage and me  
> having
> to find a 9-track to use.  Or, better yet, just zip (tar/gzip) the  
> whole
> thing up and post it online so everyone can be see what your  
> contributions
> were.  No wonder I couldn't find you doing an online BSD search (
> http://lists.hu.freebsd.org/bsd/cgi/search.cgi), we are talking  
> about code
> written when MSDOS 1.0 was first released!
>
> All those references you gave to Linux running on large mainframes as
> examples of how inefficient Linux is?!  I think anyone with a little  
> common
> sense would realize that most people with the money to run a large IBM
> mainframe are not fools, and picked the most EFFICIENT operating  
> system to
> run their application!  Your example (http://www.hoplon.com/content/view/1 
> )
> is classic, and though a bit short on details, certainly indicates  
> that
> people that want to maximize their use of expensive mainframe  
> hardware use
> Linux to run their applications on!  LOL!
>
> Paul, if nothing else you are amusing!  Well, at least I think so.
>
> Everyone else, please give Paul a hand here!  I know the OS  
> expertise is
> getting thinner every year here, but someone out there should be  
> able to
> point out a mistake or two of mine, or at least help Paul out with his
> contention that the Linux kernel is HORRIBLY inefficient.  There  
> must be at
> least some little used corner of the kernel that is noticibly  
> inefficient.
> Though not technically part of the kernel, I know there are some  
> inefficient
> drivers that could use some attention, though not sure they could  
> qualify as
> HORRIBLY inefficient, but you don't know until you analyze the code.
>
> Peter
>
> * To join/leave the list, search archives, change list settings, *
> * etc., please visit http://raven.utc.edu/archives/hp3000-l.html *

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *