Ellison: 'There'll be nothing left of IBM once I'm done' Oracle has pulled the rug out from under Hewlett-Packard's Intel's Itanium processor by yanking support of its database, middleware, and application software on future "Poulson" and "Kittson" Itaniums. It looks as though Larry Ellison wants to take on IBM in microprocessors for data center systems, man-to-man, head-to-head. "I …
Could someone explain why Larry is an asshole for wanting to compete with IBM? I mean, when IBM trash talked HP and Sun some time ago, no one here complained. Instead, they rejoiced and agreed and talked about how bad HP and Sun was. Why are there lot of complaints when Oracle are doing the same thing to IBM, as IBM does to others? I dont get it?
When IBM does something to Sun or HP - that is great and Sun and HP should be killed!
When someone does the same thing to IBM - he is an asshole.
Why?
It you don't think that everyone and the barber's cat haven't taken every opportunity to mention how much of a jerk Ellison is, you simply haven't been paying attention. Even longtime Oracle users will happily jump on the pile.
No. Ellison has always been an asshole.
Recognizing this is by no means new or "trendy".
It isn't *what* he says, it is *how* he says it. He is one of those utter shitheads who could wish you happy birthday and make you want to drown him in a barrel of cat piss. For that reason alone, I believe that he is the wrong person to be heading up a company that really needs to be looking at it how it is perceived by the public.
@Kebabbert, "Could someone explain why Larry is an asshole for wanting to compete with IBM"
Larry is an asshole. Its not about who he is competing with today, its about how he behaves all the time. He behaves like a billionaire version of a thug. He treats business more like Fight Club. He is the 5th richest person in the world, so he isn't short of money. Yet for him, business is less about money and more about the need for him to keep fighting and winning battles. He is behaving like a billionaire version of a thug.
He doesn't need to be so ruthless. He has made it, he's one of the richest people in the world FFS. But no, he wants to behave like a thug endlessly seeking to bully and control others and it wouldn't at all surprise me if he was loving every minute of it, because lets face it, if he wasn't loving it, he doesn't need to do it, because its not as if he needs the money. So I can't help thinking, what a sad narcissistic bastard (as in someone with a Narcissistic Personality Disorder and they can be evil bastards). Why doesn't he just do the world a favor and retire and enjoy his money, rather than endlessly behaving like a relentless fight seeking thug. :(
Well, let me put it this way
"IBM is one of the largest companies on the market. There is no need for IBM to act like bastards, using foul play and FUD all the time. IBM should enjoy it's position as a key player, instead of acting like a Enterprise version of a thug".
Can I be much clearer? Why is it ok when IBM does like this, but it is not ok when Larry does like this? IBM has been on the market for a longer time than Larry, and IBM used FUD even before Larry was born (not quite, but).
So again, when IBM acts like this - everyone rejoices. When Larry act like this, everyone gets furious. Hey, I once just posted an x86 benchmark in a discussion of POWER vs x86 and an well known IBM supporter got really upset, claiming that I "insulted his intelligence" for posting a single benchmark. I dont get it. When Larry is doing something, everyone gets upset and furious. When IBM FUDs everyone is happy and cheers and fill these posts with FUD. IBM attacks everyone and everything. I remember all the IBM FUD in the thread about Itanium.
I dont get it. Is there one rule for IBM, and other rules for the rest?
Just asking. I don't recall having heard, and he might be too cheap.
However, the businessmen who do invest in the professional politicians are the worst ones. The large (probably vast) majority of good, honest, ethical businessmen are just willing to play by the rules of the game. It's the other businessmen who want to game the system--and the result is laws that encourage or even require companies to become evil.
Maybe Ellison will "win" on those terms. I'm willing to consider the possibility that Oracle is distinctly more evil than IBM.
"...I'm willing to consider the possibility that Oracle is distinctly more evil than IBM..."
You must be kidding. IBM has always been the big Evil company, until MS took all attention. But IBM has never stopped being Evil, IBM continued.
You maybe dont know that IBM is the company that popularized and systematically used FUD? Read the FUD article on wikipedia and see. And also, you can see all the IBM supporters FUD a lot in almost every thread. Why are there no Oracle FUD in the comments here? No, we only see IBMers attack HP, Sun and now Oracle. And best of all, most of these attacks are "anonymous". Real cowards are what IBMers are.
With such a master spreading FUD, the pupils are the same. "More evil than IBM" - what a joke.
"If you think companies do a lot of arithmetic, "
Uuuh , yeah! What does he think banking systems do - send the sums to special maths pixies to work out with pencil and paper?? There is plenty of integer arithmetic going on specifically because floating point can cause equality test issues and other inaccuracies.
In the banking world, the problem is not doing fast integer arithmetic - that is easy to do. Anyone can update accounts with the new salary fast. There are other problems, such as quickly fetching enough data to feed the cpu.
cpus have fast integer performance enough to do banking workloads.
I can't understand why Kebabbert got downvoted for his/her comment. Maybe the truth hurts.
One of the defining characteristics of big iron is the throughput it can handle, I have worked on a Parallel Sysplexs of 14 mainframes that was quite happy to chug through 30 million transactions in slightly less than an hour. On two other separate sites I have also seen DB2 database tables with over 25 Billion records.
Mainframes do this by having channel controllers to control I/O devices leaving the CPU free to do the unsexy things like adding 8p to your bank accounts accumulated charges because you used your bank card in an ATM.
It’s horses for courses; I can’t imagine a mainframe doing complex ray-tracing calculations to create the reflections on the silver surfers body, that’s not what it is designed for, it was designed for data processing and up-time reliability, and on that it certainly beats the crap out our latest oracle based toy which has to be rebooted regularly because it is leaking memory.
Sadly for LPOD, it is not sufficient for oracle to win, but the opposition must be totally vanquished, however, the mainframe and Mark Twain have one thing in common, its the statement “The reports of my death are greatly exaggerated”
Mainframes, still kicking ASCII
"...I can't understand why Kebabbert got downvoted for his/her comment. Maybe the truth hurts..."
This down voting is because there are lot of IBM supporters here, and as we all know, IBM and IBM supporters use foul play and FUD. It doesnt matter what I write, the IBM supporters down vote it. I once asked a simple question not related to IBM, and guess what? Even that single question got lot of down votes!
And when I ask why lot of people here are mad at Larry Ellison, I draw even more furious anger from the IBMers here. They are worse than cockroaches, actually. Everywhere, spewing out the FUD and attacking everyone that questions IBMs foul play. :o)
.
But we all agree on that Mainframe have better I/O than any other server. This is a well known fact, and nothing to dispute. Just as you explain.
I'm an IBMer as well.....
As far as I am concerend <insert derogatary name for supporter of particular platform> all sprerad FUD and <insert name of company> engage in some sort of foul play.
Here's another Mark Twain quote for you "All generalizations are false... including this one".
They offered us one for $18,000 to replace a bunch of old sun hardware that costs us $1,000 per machine when they were new and would work fine if they didn't have a bug in their hardware that keeps them from working with drivers bigger than 128 gig. We already found a solution to fix the fans. Their new box would have taken more power too but might have been faster but I don't need faster.
And so does crystal meth. :-)
Seriously, do you expect customers to keep tossing out $$$ for faster and fast performance when they already have fast enough performance? That is to say, a potential customer who just built out a DW solution isn't going to need to re-invest in that solution, (sans maintenance), for at least 3-5 years.
Ellison is preaching to the customers in IBM's 'white space'.
But outside of that... All Ellison really has to do is to sit back and watch IBM crumble from rusting our from the inside. IBM has already cut as much fat as they could by offshoring back room jobs to lower cost countries. They also have off shored most of their delivery staff too. This goes only so far before their customers either negotiate IBM in to oblivion or revolt and join IBM's 'white space'.
Sure it will take some time. But its happening.
Exactly. What are we calling Java here? Is it the JVM?
If you have a Java compiler that produces native code, I imagine you could get within a factor of two or three of C++, just as you can with most other languages. If instead you are targeting the JVM, then you obviously aren't concerned with performance, so neither you nor I care how close you get.
That's fine. For many applications, performance isn't important. However, when it is, you don't target some intermediate ISA, whether that be the JVM, the CLR, or OS/400. There's a reason why all the key server applications for both Linux and Windows, as well as the kernels, are still compiled as native code and it isn't the need to hit hardware because 90% of even kernel mode code doesn't actually get to touch hardware these days.
Look at this! I am down voted again! Hilarious!
Research shows I am correct on this. My friend from my uni where a cofounder of JRockit, before BEA bought it, before Oracle bought it. And adaptive optimizing compilers ARE faster in theory.
Even researchers supports me on this. But still I get down voted. Very funny. It shows the desperation of the IBM supporters. Attacking everything I write. No matter what. And trying to discredit me and spew out FUD about me, about Oracle, about HP and Itanium, about everything.
Your performance comments about the JVM are a decade out of date. The optimizing JIT compilers in the mainstream JVMs are now very good, and JVM performance for many applications quickly converges on the performance of similar implementations in compiled-to-native languages.
Of course, the execution environment is rarely the dominating factor anyway. JVM performance will only be relevant to CPU-bound portions of applications in the first place. And within those, performance is often limited primarily by poor use of resources, inefficient algorithms and data representations, and bugs.
A few months back I was taking a course on natural language processing and had to write a part-of-speech tagger. This is a largely CPU-bound exercise (the dataset was relatively small, some tens of thousands of sentences) with a lot of floating-point calculation (basically an HMM driven by n-gram probabilities with Katz backoff). My Java implementation, which only included basic optimizations like memoization, averaged 1ms per sentence on commodity x86 hardware. A fellow student wrote his in C++ and had worse performance - by a couple of orders of magnitude.
Does that mean Java is 100 times faster than C++? Of course it doesn't. But similar sweeping generalizations about Java performance are equally vapid.
I admire the corporate ambition on display. But I think that there's a lot more to IBM than just integer performance. I think that Ellison is underestimating IBM and their appreciation of their customers' needs.
There's plenty of hardware out there (Sparc, Itanium, x64) that should theoretically mean that IBM's hardware offering is lacking in appeal in one way or other (performance, cost, or whatever). But apparently that's not reflected in IBM's sales figures.
I think that IBM are actually quite subtle as to what they put in to their hardware designs. My favourite example is the decimal arithematic hardware acceleration on the POWER processors. That's absolutely perfect for massive banking applications having to process international transactions. As Boltar rightly points out in the post above, ordinary floating point is not accurate enough. Almost no one outside that niche knows that it's in there. But it shows that IBM have really thought about banking applications all the way down to the CPU design. And guess what - IBM sell to a *lot* of banks and financial processing outfits.
Whether or not Ellison understands that point I don't know, but it is important. IBM clearly has means of offering cost effective systems to their customers in ways where individual benchmarks are irrelevant. The customer ultimately cares about only service-per-dollar. This has allowed IBM to sell a surprising amount of mainframe gear for many decades now. So much so that everyone seems to have given up saying that the mainframe is dead.
Anyway, whilst Ellison may witter on about Java, there's a shed load of COBOL out there, new and old.
Having said that, I do like what they've done with SPARC.
"...My favourite example is the decimal arithematic hardware acceleration on the POWER processors. That's absolutely perfect for massive banking applications having to process international transactions. As Boltar rightly points out in the post above, ordinary floating point is not accurate enough. Almost no one outside that niche knows that it's in there. But it shows that IBM have really thought about banking applications all the way down to the CPU design...."
Again, if you are doing it right, you never use floating numbers in finance. Every calculation is done with integers, and you keep track of the number of decimals separately. No rounding will occur. No floating numbers are needed. As I said, I work in a large finance company.
.
"...And guess what - IBM sell to a *lot* of banks and financial processing outfits..."
I agree that IBM sell lot of gear to banks. The banking world relies heavily on Mainframes. Banking world updates an account with the new salary, decrease the account with rent, mortgages, etc. Those are trivial calculations, done in COBOL on Mainframes. Not very sexy. I see old dusty rooms with old men sitting with monochrome terminals doing COBOL.
In the finance world, Mainframes are never used - they are too slow. Here we typically use Linux/Solaris and C++ and do High Frequency Trading, algo trading, quant math, risk analysis etc - now THAT is sexy! Banking world is boring. Finance is cool. I see sky scrapers, suits, MBAs, Quants, Traders, HFT, algorithms, Hedge Funds, Wall Street, etc.
Traditionally, Solaris has been used in Finance and Telcos. I have never seen an IBM system in finance. I work in finance, not in banking. It is always Linux or Solaris.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
