Oracle sucks major horse ass no matter which platform it’s used on. There’s a reason the story of the emperor’s new clothes exists….

i have the new unix in mt mac computer , i can tell to all microsoft fan – linux is alaive

I would point the finger at DELL before linux. Linux 2.6 kernel is rock solid. Dell servers? Issues – many issues. We manage 200 of them. and about 40 Compaq and 100 IBM.

I’ll take a Compaq (DEC) over a Dell any day. Unfortunately they cost significantly more.

I’ll be unable to reply in detail, but would briefly make these points:

“Nowhere does it say that Free Software costs nothing to use. Only the most clueless computing newbie would think that.”

Apparently not:

“* Now I have a position where I travel even more and have witnessed how ubiquitous F/OSS is and how many shops have little proprietary software and nearly ALL F/OSS. Not 100% but the trend is definitely happening for the reason I stated above – mostly cost.”

“And learning how to write good code, IMV, comes from reading manuals (not some other goober’s perspective on how to do things) or better yet, pouring your eyes over someone else’s code.”

As I said above, I can’t possibly agree more, and is how I’ve learned programming. I took courses in local Community College (didn’t get degree because of required Speech class), read manuals and looked at others code. That was in mid to late 70’s, so excuse me if I point how ludicrous it is to imply, let alone state, that F/OSS is of any significance whatsoever in this.

“Yeah, those of you whose UNIX background go far back, I can see how F/OSS appears to be an inferior reincarnation.”

Which it is.

“But you’re not focusing on the big picture – ”

I happen to be a “big picture” guy, Naum. You ain’t gettin it a bit. For example:

“Your quoted numbers, no doubt, are accurate. But that still doesn’t mean squat – tests can be fudged to show anything. … Does that mean MVS > Linux on OS390? Or is it just the unique set of factors there on that implementation? Or supposing it is much better, is it less costly? Or maybe it is, … … many, many variables, cannot just give me a benchmark and tell me A is better than B… ”

Well, anybody that makes a judgment just based on “the speeds and feeds” is a fool, no doubt, but people are swayed by funny things. People get corn-fused, and make judgments about technology that represent their political views, rather than what is most effective for the enterprise, quite frequently. Poop happens. And since every entity IS unique, it’s real easy to work backwards from conclusion to fact: “Well, in OUR org, THIS gear works THE BEST ’cause [fill in the meme].”

“What is rather amazing is how much people talk about Linux needing less handholding than Windows (or the contrary depending on the interest group), but no one really asking companies: How many servers do you maintain for each OS, and what FTEs (people) do you assign to them?”

Even more incredible is how much people talk about how many servers they can run, rather than how few to do the same amount of work. Bowoggles the imagination.

And Henri, if you can consistently say this on ANY gear “ok, I can guarantee 100% successful implementation for price $P1 and time T1″… well, I’m sorry if I’m skeptical. (A Teacher got me some work while I was a student, so..) I’ve been doing this stuff professionally since about ‘77 and I’ve never guaranteed successful implementation of any project of any size. Smaller stuff, yeah… In spite of this lack, I’ve done a fair number of large and small projects, btw.

Mebbe laterzzzzz…:-D

The thread here has diverted from the gist of the article and I guess my thoughts on the article weren’t lucidly represented. I just found it astonishing that these respected programmers (and including Philip’s commentary) “can’t see the forest for the trees”… …bemoaning the fact that F/OSS is primarily a re-incantation of UNIX (with a Windows GUI on top…) while remaining totally oblivious to a revolution that is occurring because of F/OSS. Disentangled from the vagaries of contemporary day-to-day computing, they drone on for a “sliver bullet” or some evolutionary breakthrough that flies in the face of the mountainous base of existing software and hardware, that increasingly inhibits the onset of any new radical transformations in the field.

Simonyi looks like he’s trying to reinvent CASE tools. Yeah, maybe he should stick to formulating a new and improved variable naming notation…

Back to the F/OSS significance – maybe it’s because I’m an old-school programmer that came late to the *nix party (around 1997 or so). For many years I worked on Burroughs, DEC and MVS systems. There was no sharing of code going on – in fact, I was lucky to get access the manuals for such systems. The first university I attended got in big trouble for making copies of the VAX manuals. And learning how to write good code, IMV, comes from reading manuals (not some other goober’s perspective on how to do things) or better yet, pouring your eyes over someone else’s code. Yeah, those of you whose UNIX background go far back, I can see how F/OSS appears to be an inferior reincarnation. But you’re not focusing on the big picture – the acceptance and widespread dissemination of F/OSS into mainstream commercial and personal computing realm.

* 10 years ago, it would be unthinkable for most Fortune 500 firms to adopt F/OSS or anything that wasn’t supplied by IBM or MS or other software vendor like Computer Associates or Serena or $insertVendorName. Unless it was a shop-brewed deal, but by this time, most programmer staffs had been slashed enough to curtail such undertakings.

* Even recent as 3 years ago I came into an AIX shop that at first, strictly forbid me from writing any Perl as they had an official policy against using F/OSS. But by the end of a year stint there, substantial portions of the systems were rewritten in Perl by me.

* Now I have a position where I travel even more and have witnessed how ubiquitous F/OSS is and how many shops have little proprietary software and nearly ALL F/OSS. Not 100% but the trend is definitely happening for the reason I stated above – mostly cost.

* F/OSS is in large part responsible for the explosion in web sites (some may discount this event as not a positive thing …). Most folks arn’t going to plug their own server in or even opt for leasing their own connected server – they’re going to go virtual host and for a cheap fee, they’ll be served with LAMP. Just take a look at the Apache dominance of the web server realm. Linux may be a bit role player on the desktop but it’s becoming king in the web server market, at least on a personal/non-profit level and even on an enterprise level for many companies.

* It’s going to play a role in the coming years on the lines of DRM and so called “trusted” computing. An alternative to the locked down platforms that new Windows platforms appear to be headed.

* Finally, the most critical aspect and the item here that eclipses all others – the immense power of “code sharing” on a global level. Basically, all have the tools at their disposal to become extraordinary programmers. Not just coders who toil for Microsoft or for blessed individuals who somehow got the access to expensive computing resources. Anyone, if so motivated, can view the code and get their fingers dirty…

JayT, as far as numbers and benchmarks go, whatever. Your quoted numbers, no doubt, are accurate. But that still doesn’t mean squat – tests can be fudged to show anything. The only relevance is how it affects the systems that I’m running. Another anecdote to end my blurb here – at a previous employer, relayed to me by friends still holding positions there, heavy IBM shop that I supported an application handling millions of transactions and terabytes of database storage has moved from MVS to Linux on OS390. CPU utilization (which for this application system, is heavily monitored and critical) has skyrocketed, and performance has declined. Does that mean MVS > Linux on OS390? Or is it just the unique set of factors there on that implementation? Or supposing it is much better, is it less costly? Or maybe it is, but IBM wants the client to ante up more money to compensate for their assuming more support in a capacity that they understand not… … many, many variables, cannot just give me a benchmark and tell me A is better than B…

JayT:
The main reason F/OSS software is written when something already exists to do the job is not to create a revolution, but to simply allow the software to be used by everyone, and potentially by people who will tweak it to use it in unforseen ways. The software is not revolutionary, it’s the license to use it that is.

And I did feel the revolution, not being old enough to have enjoyed the free exchange of ideas in the software industry of the 60s and 70s (those were the true revolutionary times). 8 years ago when we started BizRate, I had to cobble together a number of proprietary solutions, thanks to some really useful scripting environments on the macintosh (Userland Frontier being the main one). When we switched to Linux in 97, it changed everything. This doesn’t mean that I couldn’t do it with other hardware/software at the time, but Linux made it simple and easy. Finally I could use the software that other people had made in ways that they never expected, but were great for ME. That’s the beauty of F/OSS.

You then state: “Then WHY market it as “free”, in the first place?”
Free software is not “free” as in costing nothing. It is “free” as in allowing you the freedom to do anything with it. Free Software is costly. Everything is. You have to install it. You have to maintain it. In the case of Free Software you may even need to modify it. You also may need support. Nowhere does it say that Free Software costs nothing to use. Only the most clueless computing newbie would think that.

But I stand by my statement that while F/OSS costs money to install and tweak properly, in most cases the TCO is massively lower than closed-source proprietary implementations. You of all people should know, with your experience, that the secret of a good implementation is in the details. With F/OSS software, you not only know what you’re getting ahead of time, if you hit a snag you can get rid of it. With non-F/OSS you’re totally out of luck and at the mercy of the vendor, and it makes the implementation failure risk dramatically higher.

There’s a simple example to illustrate this point: you’re tasked to implement a system. You go to F/OSS software that may have 90% of the functionality, You implement a pilot, you cost out the remaining 10%, then you say to your manager: “ok, I can guarantee 100% successful implementation for price $P1 and time T1″. Now assume you go for a proprietary solution. Assuming the vendor doesn’t lie about the functionality in his software, and you manage to suck out of him a free or cheap pilot, you may expect to get 100% of the functionality. The only thing you can tell your manager is: “I think this will work, and I’m 90% confident that we’ll have a successful implementation (barring unforseen problems in the details), and it will cost $P2 and should take T2 time”.

Well, the more complex the system to build, and the more external dependencies with other existing system, the worse off you will be with a proprietary solution. Your confidence level will drop, and your implementation will be a lot more likely to fail. And that is the true cost of proprietary closed-source software.

Then you want to compare Linux to OS/400. OS/400 is a great OS. I’ve barely used it but I have a lot of respect for it. Now find me the staff to build software on top of it.
What is rather amazing is how much people talk about Linux needing less handholding than Windows (or the contrary depending on the interest group), but no one really asking companies: How many servers do you maintain for each OS, and what FTEs (people) do you assign to them?

I only know from my own experience, and I can safely say that I have about 1 FTE per 100 high-performance Linux servers. This includes setup, monitoring, maintenance, networking, work on the co-location facility, UPSes, etc… What is your experience with Linux? Windows? OS/400? (by the way, Solaris is even better than Linux on the FTEs, but it’s a “cheat” because we use Sun support on those, so overall it’s about a wash)

Next, you say: “How many can afford staff to custom-develop OS components??”
Well, anybody who needs something that can’t be done, or can only be done at other prohibitive cost. Think of F/OSS simply as the opportunity cost when you are pricing some proprietary solution: how much would it cost me if I got some F/OSS and customized it? That’s all you need to do.
As I said above, you’re going to pay for what functionality you need, either by buying it or by making it. If you’re lucky, someone has already built all the functionality you need in a F/OSS package, and you won’t have to make it. In most cases, you’ll probably have to do 20% of the work, 80% of it being already available.

BizRate can “afford” staff to custom-develop OS components because we found out that there’s a lot of value in the software we’re creating. Many more companies should think about it and look at what value they can create internally by building what they truly need, not by compromising and patching together existing off-the-shelf components.

Naum, when you did your comparison of Linux and how superior the TCO is to OS/400, did you spec out Linux on the 400? The iSeries won best-of-show there 2 years ago.

And btw, Naum, IF you actually did some research in this, did you scope out Linux native on an LPAR, or was that Linux running on a 400 PC co-processor?? And finally btw, if the latter, was that internally connected on the system bus, or external via HSL… I’d be curious as to some of the details of this comparison, as I actually WOULD like to learn some more on this stuff.

And, when you say “I’d rather pay to keep a talented staff aboard and let them hack away on the open source stuff then shell out big bucks to IBM consultants and all the exorbitant licensing fees” I’m with ya there. However, the question is to how much you can fork out the nose on talented staff for most businesses I’m familiar with, which are not largely able to pay ANY money for IBM consultants. You must be in a lucky situation where you work, to have that kind of money.

First of all, Mr. Gabbour, when it comes to stupidity we may very well be a close match, but when it comes to ignorance I’m afraid you still have me beat by a long mile. There’s been bazillions of companies that produced great tech that didn’t survive in the market very long. Yeah, I’ve seen most-a the flashy products come and go.

I’m no expert on 5L, in case I wasn’t clear. But I do know a 400, and with (what?) 16 processors it was able to smoke all the other’s in VolanoMark (Java) Benchmark a few years back. Tops… Iow, it’s not HOW MANY processors you have or HOW MANY GHz you have, right? It’s what you can do with them. And haven’t kept up on the feeds and speeds, but there are plenty of (slightly-unsubstantiated) rumors that IBM will NOT PUBLISH the full stats on a 400, because it makes their high-end mainframes that cost 10X as much look pretty wobbly.

And you mistake me entirely, fool, because I’m not smart and have only a high-school degree at that… I’m just old and grizzled, so happened to see so much of this so-called “revolutionary new Great THINGS!” come out a decade or two back and fail. However, being old and grizzled allows me to see the reality of things better, and not be so EASILY impressed by the flash and sizzle that most are.

OTOH, in the late 80’s or early 90’s I had my IBM CE (Customer Engineer, the hardware gearheads) give me a call. Said my hard drive was a li’l flakey, but percentages were that it wouldn’t be going down any time soon, so we could schedule a replacement at my convenience. (That was old days, prior to hot-swap.) I’d seen the message on QSYSOPR *MSGQ, so wasn’t surprised by that, but I’d only HEARD-a folks getting called by their CE’s like this, prior to that.

Iow, I’m not smart so much as I’ve been lucky to be working on (a LOTta aspects of) “tomorrow’s technology” a decade ago.

Any-a y’all heard of Grid or Autonomous Computing and all that…?? No, I don’t claim to have INVENTED it, just worked on it’s precursors for many-a year.

Finally, Mr. Gabbour, I’m sorry but I haven’t heard-a F & H so plead ignorant. (Ah well.) But I’ve worked on 400’s in companies from measely $10M to Fortune-30, and recall how (at one time anyway) Allstate bought 10,000 of ‘em to put in their offices with NO staff whatsover to hand-hold them… I’m afraid neither you nor anyone else is gonna be able to do better than that, as far as versatility and reliability and cost-effective… If you stick to facts rather than anecdotal evidence that primarily jes goes ta-show the exception-to-the-rule.

You’ve been in this industry for 20 years JayT, and the best you know is an IBM server? 5L seems to only scale up to 32 processors. In comparison, the CM-2 could ship with 64,000. Not bad for something people like Feynman and Hillis happily worked on. Sure, it went bankrupt after 11 years, after the supercomputer market collapse, but I’m sure a grizzled old veteran like you is used to that. I always have to go back in time to learn about computing.

You can do better than that, please try again.

Sure, I could be less ignorant and stupid as you claim, I notice it often… but that would mean being only as smart as you. ;P

>>That was my point about your views, Naum. Do you even know what CPF and an i400 is?!?

Yes I do, but it’s not germane to the subject matter here. So you’re an IBM fanboi … is all you’ve displayed.

But I will say flat out IMV that Linux is superior to OS/400 in overall terms of TCO … IBM support is very expensive, I’d rather pay to keep a talented staff aboard and let them hack away on the open source stuff then shell out big bucks to IBM consultants and all the exorbitant licensing fees.

Of course, it doesn’t matter what I think or what you think. Fact is that the free software/open source stuff is becoming much more prevalent as corporate entities are seeing that it makes no sense to plunk down big money when there are free alternatives that are just as good or good enough.

Mr. Asseily,

I thought I recalled your name from the SM Community, which I infested briefly a couple/3 years back…;-D

I read every word with great interest (prior to fact-checking the above, btw). I just don’t happen to have the same experience in these regards:

“- Free Software is not about building revolutionary products. It’s about building products that can be used in revolutionary ways.”

I’m sorry, but today my tolerance for old lame meme’s is running low. And, in almost each and every case, whenEVER I see folks heading towards “revolutionary”, (in IT or anywhere else,) I see them re-inventing the wheel, and generally doing it poorly. It stands to reason, right?

“Being able to freely change what you don’t like about a piece of software leads to massive cost savings.”

Not always, as you yourself note in what follows.

“JayT, this is what you miss in your statements that costs are huge. I agree that overall it may well cost MORE to build the core of a piece of open source/free software,

Then WHY market it as “free”, in the first place?!? (Rhetorical question, obviously. Because stupid people actually believe there is such-a thing as a free lunch.)

..but it costs SIGNIFICANTLY LESS to implement as soon as the implementation is non-trivial, i.e. when the implementation is in a heavily inter-related system. …but that’s isn’t as great as the value you get out of Apache, where you can tweak everything to your heart’s content.”

Firstly, this would contradict the errant meme because MOST businesses do not have the LUXURY and HIGH-MARGINS to economically support a bunch of dweebs tweaking OS components to their “heart’s content”.

And secondly, or redundantly or whatever.. That is the PROBLEM. Company’s are getting bled dry by geeks/dweebs/nerds/whazthediff who wanna play like they are systems programmers. VERY expensive meme, that one, and many of these guys are NOT very GOOD systems programmers, at that. Total Cost of Ownership-Wise, those expenses add up, even if they are just buried or hidden in IT budgets… It still adds up to expensive.

“- Regarding Linux reliability. Mr. Greenspun seems to have made a joking poke at Linux that was blown out of proportion.

Quite the contrary. It was never blown INTO the proportion it deserved:

“as I understand it one of the guys has essentially had to move into the colocation cage to keep poking at Linux.”

Yeah, those server farms are so CHEAP, unless you add the cost of an increasing (and increasingly expensive) head-count. Which most IT managers DON’T ACCURATELY KNOW, let alone figure in. And then they really don’t have adequate figures to do a COMPARISON against, other than what they’ve seen or heard of Windows shops.

“In my experience, Linux is significantly more reliable than Windows.”

And how does it compare to OS/400??? It doesn’t. How much time, just for one single for-instance, does the paid staff spend keeping up on security updates?? Can either of these OSs verifiably claim they have NEVER had a virus in 25 years experience??? I’ve never heard the claim. Dunno about Solaris, (my bad,) but never heard it has never had a virus.

“Again, assuming that you value speed so highly that it’ll be worth your building a customized engine, …”

See, that’s a very valid point.

But one that doesn’t generalize very well to very many businesses a-TALL. How many can afford staff to custom-develop OS components??

Well all too many do, but they can’t afford it like BizRate can. (Or they’re affording it by shipping jobs from America and Europe elsewhere, but thaz a whole ‘nuther story I don’t have time ta-get into.)