Re: IBM's realised file systems are key.

"...What is needed is efficient algorithm that stores more data,..."

What do you mean? Can you explain?

ZFS stores 2^128 data. That is more than mankind have produced, or will ever produce. To store that kind of data, requires moving electrons so much that the energy needed to move electrons, would boil away all water on earth. If you try to store 2^128 bits on 3TB disks, you need as many disks as 10 moons weigh. And the moon weighs a lot.

So why do we need to store more data? Maybe I misunderstood you?

Re: Wow, HP realy are slinging there own mud about

@PXG

"...IBM was never going to buy Sun and anybody in the industry could of seen that was the case, IBM dont like being in a position of being a monoply due to historical bad press..."

What? IBM loves monopoly, dont you know that? IBM fights very hard against any company that tries to break IBM monopoly on the Mainframes. IBM is sentenced to billion dollar fines, or buys the company. In case you didnt knew, IBM earns lot of money from their Mainframes. It is really big business. In practice, IBM has monopoly on Mainframes and fights to uphold it.

Every company strives to a big market share as possible. The best is monopoly, and the ultimate goal. That is the reason we have laws against monopoly - it is too lucrative. IBM not wanting monopoly? You should read basic courses in business or economy.

.

.

Of course IBM wanted to buy Sun. They were far in negotations when Oracle snapped Sun in front of IBM. IBM would have killed Solaris and SPARC, and offered a cheap migration path to POWER/AIX. Solaris is the most common Unix out there, so IBM would have a big market. Then only HP-UX would have been left. And IBM would soon have reached monopoly on Unix too. IBM could then increase prices on Unix as they liked, and have monopoly on both Mainframes and on Unix servers. Left would Linux be, for lowend. IBM would have monopoly on all highend servers. And highend is high margin. Low end is low margin. Golden opportunity. Very very very lucrative. IBM must have been really pissed when Oracle snatched Sun.

You are totally off, believing no company wants to increase its market share as much as possible. AIX is not most common Unix, but IBM wants it to be, no matter what you think. Monopoly is the ultimate goal. Or close to monopoly.

Re: IBM's realised file systems are key.

"...With huge and increasing amounts of data, the simple filing systems of Linux/Mac/Windows are no longer adequate. Advanced database filing systems where files have extended metadata for both the user and the file data are sorely needed...."

Ok, and what advantage does the metadata you talk of, give us? Why do we need it? You only say it is needed, but why? Please enlighten us? :o)

.

.

How does this GPFS compare to Lustre + ZFS? The new coming IBM supercomputer at 20Teraflops will use Lustre + ZFS. It will have 55 Petabyte of data, and 0.5-1TB/sec bandwidth. They are porting ZFS to Linux Lustre, because ext does not scale.

http://zfsonlinux.org/docs/LUG11_ZFS_on_Linux_for_Lustre.pdf

http://www.youtube.com/watch?v=c5ASf53v4lI

Nor does ext protect against data corruption. All hard disks have checksums to detect and correct data corruption, but that is not enough. Every once in a while, the disks will encounter random bit flips that the checksums can not correct. Even worse, some of the errors are not even detectable by the disk. The only solution is to use End-To-End checksums, which ZFS does. Thus, ZFS detects and protects against data corruption. There is research on data corruption and ext / ntfs / XFS / JFS / etc here (research shows that ZFS does protect your data):

http://en.wikipedia.org/wiki/ZFS#Data_Integrity

Re: A large Investment bank I talked to

@Jesper

Have yor studied the official roadmap I linked to, SPARC will double the performance every other year? Now, if Oracle succeeds in keeping that goal, then do you agree that the goal is more aggressive than any other cpu?

There is no roadmap for a cpu vendor which doubles performance every other year. I mean, Ivy bridge was 20% faster or so. And Haswell might be 20% faster. It is not like Intel is doubling performance like SPARC does.

If the SPARC roadmap is true, then SPARC gets the most improvements in performance, more than any other cpu vendor. And already today, T4 beats POWER7 in some benches. The eight socket T5 servers will give four times the performance the T4-4 servers.

Do you know of any other cpu vendor that doubles performance every other year? No?

Re: A large Investment bank I talked to

"...Solaris 10 branded containers on Solaris 11 isn't viable for some applications due to vendor support..."

LDOMs will not do?

Re: Hmm...

"....As far as calming down worried users.... As Power has never had this large a percentage of the Unix market share in the history of Power servers, it doesn't seem like there are too many people wringing their hands...."

Something similar said the Dinosaurs just before they were wiped out. Everything looked good on the surface, but a deeper analysis shows there are fundamental problems.

For the first time, x86 is seriously threatening POWER: performance and RAS. No one can just shove this fact under the carpet and pretend it is not true. IBM needs to do something radical. A cautious General prepares for the worst case. IBM should do that too. Maybe start to diversify and not rely to heavily on proprietary (soon inferior) IBM tech such as POWER and Mainframes. Maybe start to offer Linux on IBM prop tech. Maybe preparing for exit.

I heard that POWER7+ is late, but it is not late, you say? Can someone confirm if it is late or not?

.

You all like when I show credible links from different IBM executives, and analyze the trend, and talk about the dark looking future of AIX and POWER, right? Again and again, right? How do you feel? Warm and cosy? How do you think we felt when you started spreading the FUD about HP and Sun and Oracle many years ago? What comes around, goes around. If you IBM supporters hadn't FUDed so aggressively earlier, no one would be interested in rubbing this in your skin. Suit yourself.

Admit it: x86 is catching up on POWER, and IBM is betting more and more on Linux. IBM starts to offer, for the first time, cheap POWER Linux only servers? Bad sign. What does that mean? That POWER has a rosy future? Hardly. IBM only does high margin business. I have talked about it for years: IBM needs to cut the price of POWER, as cheap x86 catches up in terms of performance. This is happening. Now. Today. My predictions are starting to come true. The next step is that when POWER prices are low enough, IBM will kill POWER and switch to x86. Probably, IBM will continue with cheap PowerPC cpus, but expensive POWER cpus will be killed.

Re: Hmm...

"...This 'new' ability to only run Linux is not new...."

No, I agree with you on this. But the new thing is that IBM now offers Linux ONLY servers. Earlier, you could run Linux in different ways on POWER, for instance virtualize Linux. That is not new. The new thing is that there are Linux only POWER servers now. Why is that? That is new.

.

.

"...Now. I'm not going to argue with the fact that AIX (along with all proprietary UNIX systems) is on the downward side of the popularity curve, and I do not think that PowerPC development is in a good place at the moment.... Sometimes I wonder whether IBM really wants to remain in the hardware business at all (products that have been sold include their printer division, their storage division, the desktop and laptop PC business, and most recently their ATM and PoS business)....

I am really not looking forward to a point where the only processor game in town is X86 derived, and that is looking like a possibility within a decade unless ARM moves upward..."

It seems that you agree that the trend shows that x86 is fast catching up on other cpus. The future looks grim for other cpus, they will be eaten. Only niche cpus can survive. Intel does not venture into every niche (yet).

Myself dont like x86, it is too buggy and bloated. But analyze the trend and think a bit. The future is clear. POWER7 is going out, IBM is preparing for exit. POWER7+ is late, maybe IBM has problems with performance. IBM needs to beat x86 severely. Just little will not do. And about POWER8 we hear nothing. Is that canned? What happens? IBM should say something to calmd down worried users? Another thing is that pessimitstic people seems to be right and correct.

Re: RE: great if all you run is Oracle

Sadly, dedup on ZFS does not really work well. For instance, you need lot of RAM, something like 1GB RAM for each TB disk. I dont know how NetApp handles dedup? Are they more efficient?

I think that we can summarize as this ZFS server from Oracle is a good step. Now these ZFS server are low-end to mid-end. But high-end still is out of reach, for instance the 140 node isilon cluster.

Thus, if you need low-end or mid-end servers, then ZFS will do. Later, ZFS will venture into the high-end segment. But that will take time. But for now, there is a new sheriff in town, and his name is ZFS.

Re: SPC-2 too

I think Oracle used 300GB disks, while NetApp used 450GB disks. Oracle also used mirroring, effectively cutting storage in half.

Anyway, it is interesting that you can buy several Oracle ZFS servers for the price of one, slower NetApp server. Something is not really correctly priced. Either Oracle should be much more expensive, or NetApp should be cheaper. Or both.

Re: SPC-2 too

I think the cost difference is interesting. NetApp is expensive for similar performance. Why is that?

Re: Performance increase is not goal of a replacement....

"...if he really wanted to kill off AIX, why haven't he done it ?..."

It's not "he" alone. Several IBM execs talk about killing off AIX in the future. Not one guy. They talk about "multi-decade time frame" and "it will not happen soon". They also say the "future is clear". IBM is betting more and more on Linux.

Do you expect POWER8 to be 50% faster than POWER7? That is not enough to keep Intel at bay. All of us can see it, x86 _will_ catch up and surpass POWER. But that will take another 5 years of rapid development. Then IBM will struggle a bit, lower prices etc, before succumbing to x86. After another 5 years of struggle and diminishing margins, IBM will kill off AIX and POWER.

Thus, nothing in the IBM links I post here, contradicts reality. And then the IBM FUDers can sit on your antique AIX POWER6 and POWER7, while the rest of the world has moved on to x86 and Linux / Solaris / FreeBSD. And there was much rejoicing. But I doubt the IBM fud will stop then, it will only shift focus...

Re: Performance increase is not goal of a replacement....

"...It's the same and the same and the same and the same and the same and the same again and again.. and it doesn't really change over the years...."

Maybe you should ask yourself WHY am I posting this? Why do you think? Is it because of the IBM people posting the same FUD again and again and again and again all the time? And it does not really change over the years.

As I have said many times, when IBM supporters stop posting their false FUD, then I will stop posting my replies. And besides, my replies are always backed up by credible links. Whereas the well known FUDer Allison Park does not show any links.

But when I post links where IBM itself says that AIX is going to be killed off, that link must be considered as credible.

Re: Garbage Comparison

@M.B

I suspect that Oracle storage offerings in > 1 million USD price range, are totally differently speced than this 0.2 million ZFS server. Dont you agree?

Re: smaller (and 18 month older) apples but 4x more of them

@W&W

I agree that Oracle using more memory is not really fair in this bench. But, on the other hand, NetApp costed a million more. If NetApp added more memory, NetApp would be even more expensive. Noone can dispute the big price difference, more RAM or not. If Oracle increases the price one million, Oracle could add much more hardware and other features.

No matter how you see it, the NetApp solution is hefty over priced compared to Oracle. That we can all agree on.

Re: Performance increase is not goal of a replacement....

@David 14

"...In most every way, [Oracle] are behind IBM hardware, and behind AIX and even RedHat capabilities in the OS. The last couple of bright spots seem to to only be DTRACE and ZFS, both of which will only appeal to a small number of business customers...."

Same old. You have just been fooled by IBM marketing and IBM FUD. Oracle is not behind IBM hardware. Let me dispel the IBM FUD for you. First, Oracle holds several world performance records, thus Oracle hardware is not slow, in fact it fastest in the world in some benches.

IBM mocked Niagara cpus, stating that the future was a 1-2 core cpu, running at 7GHz or faster. Because "databases runs best on strong cores. Thus, having many slower cores was just too dumb. Sun realized that GHz race was leading nowhere, and instead switched to many cores. Back at that time, 8 cores was crazy. POWER6 ran at 5GHz, and had two cores. After IBMs rant about how Niagara was bad, it was expected that POWER7 would have 1-2 cores, running at 7GHz or faster. Well, look at POWER7 today. Clock speed has been significantly lowered, and IBM has followed the many core race instead. Just like Niagara pionereed. That is just too dumb of IBM. When Sun/Oracle does something, it is bad, but when IBM copies several years later, it is the best thing sinced sliced bread.

Have you heard about the "worlds fastest cpu IBM z196"? Well, as I have proved with links to Mainframe experts, the cpu is slower than a decent x86 cpu. So how can a Mainframe sporting 24 slow z196 cpus, replace 1.500 x86 servers? It is just classical IBM marketing and FUD.

Solaris is far ahead of AIX. The most innovative OS today, is Solaris. Everybody is copying or porting DTrace, ZFS, Zones, Crossbow, etc. Have you heard about IBM Probevue? That is a DTrace copy. Have you heard about AIX WPAR? It is a copy of Solaris Zones. Sure, IBM did lot of innovation like LPARs (that Solaris has copied), but that was decades ago. And IBM does not have anything like ZFS, that protects your data on disk against data corruption. Development pace of AIX has slowed down, nothing innovative comes from AIX anymore. IBM is shifting resources to Linux.

Also, IBM AIX just recently had problems to scale to P795 and handle as few as 32 cpus. IBM had to rewrite AIX for it to handle those few cpus. In a three years, Oracle will release a 16.384 thread Solaris server with 64TB RAM. IBM has no chance to catch up on that massive scalability. That many threads sounds sick now, but 8-cores sounded sick back then.

And maybe you missed it, but IBM is going to kill off AIX. I am not making this up, it comes straight from IBM themselves:

http://www.zdnet.co.uk/news/application-development/2003/01/29/ibm-linux-will-replace-aix-2129537/

"The day is approaching when Linux is likely to replace IBM's version of Unix, the company's top software executive said, an indication that the upstart operating system's stature is rising within Big Blue."

Let us face it.

-POWER6 was several times faster than x86, and costed 5-10x more

-POWER7 is 10% faster than x86 and costs only 3x more

-POWER8 will have a hard time catching up on x86 perfomance wise, which means IBM needs to lower prices further. Thus, POWER will be low margin business. IBM only does high margin business, which means POWER will be killed off, too.

Thus, AIX and POWER will be killed off. So I advice you to not really buy the IBM marketing and FUD right off.

HP-UX has a dark future. AIX will be killed. Left is Solaris, and it also runs on x86. And it is open. Thus, only Linux and Solaris will be left. I suggest you shift from AIX to Linux in your future studies.

Re: I smell something funny

@J.T

Seriously, what are you talking about? You obviously have no clue about ZFS and what is does. I suggest you read more on ZFS before writing weird things. Or are you just trying to spread FUD?

.

"...ZFS appliances give you less than a 100TB usable due to everything having to be RAID 1 and a lack of hardware RAID protection making them use a lot of empty space for sparing (plus a slightly higher amount of ZFS overhead)..."

What less than 100TB? There are PetaByte ZFS servers out there. Wrong.

And what do you mean with "everything having to be RAID 1"? ZFS has no RAID 1. ZFS has raidz1 (corresponds to raid-5) and raidz2 (corresponds to raid-6) and raidz3 (corresponds to none, it allows 3 disks to fail which is unique). Wrong again.

What do you mean with "lack of hardware raid protection"? Dont you know that hardware raid is quite unsafe, and that ZFS provides a superior protection? Dont you know that there are research on this? Dont you know that hardware raid might corrupt your data, without you ever knowing it? Dont you know that ZFS is built to detect data corruption so ZFS is also safer from a data corruption view point? Have you missed the research papers on data corruption? Dont you know that hardware raid is susceptible to "write hole error", but ZFS is built to eliminate that design flaw? Dont you? Wrong again.

.

"...So basically, you aren't going to get that many backups, so you cannot do things like keep historical backups...."

This is just so wrong it is not even funny. ZFS has snapshots because it is CoW. Each time you make a snapshot, every change is written to a new place on the disk, but all old data is still left intact. This way you can have many snapshots (costing nothing, only the changes are recorded) going back and forth in time. Just like Apple OS X Time Machine. So you can keep historical backups.

Do you work for NetApp or EMC, trying to FUD about ZFS? Or are you just ignorant, claiming weird things without even nothing anything about ZFS?

Re: Forks, Spoons and Knives.

Illumos is an fully open sourced version of the Solaris kernel. It is similar to Linux. There others that provide distros built upon Illumos. For instance the storage Enterprise distro Nexenta that is starting to replace NetApp and EMC out in the companies. Or Illumian (which is Illumos + Debian package system). OpenIndiana (which is similar to Solaris 11). Schillix. Belenix. and dozens of other distros.

Re: Oracle extends Linux support to 10 years

"...What exactly has Oracle contributed to Linux, and specifically under the GPL license?..."

Have you heard about Btrfs? Or DTrace (which is coming to Linux)? Or Zones (which is also coming to Linux - I think)?

Re: Re: Re: Oracle vs. Red Hat?

"...You know that raw clock speed is absolutely not the point of a mainframe, right? A good analogy is a Porsche compared with a freight train...i.e. not really comparable! For all the criticism of mainframes, for many workloads they are untouchable, ..."

I agree with your post. What you say is true. There are workloads that Mainframes are better than anything else. Absolutely. I have never ever denied this.

BUT, if we speak of the Mainframe cpu, that IBM dubs "worlds fastest", they are slower than a decent x86 cpu. Of course the entire system can be fast on some workloads - I have never ever denied this. But let us reason a bit. How can a slow cpu, replace a faster cpu? I mean, IBM claims that 24 of these slow Mainframe cpus, can replace the workload of 1.500 x86 servers. Does it sound reasonable?

Re: Re: Re: Oracle vs. Red Hat?

"...Did you notice the lack of logic in your post ?

IBM mainframes poor but cost a lot - Oracle DB great but cost a lot..."

No, I did not notice the lack of logic in my post. Can you point it out?

There is nothing illogical with my post. Reread it. Mainframes cost much because of vendor lock in. IBM has in practice, a monopoly on Mainframes. Thus, IBM can charge whatever they want. No matter how slow the Mainframe cpus are. Sure, a Mainframe has good I/O (because of lot of help cpus) but if we only speak of the cpus, the cpus are slower than a decent x86. This is a fact. Nothing illogical here. It seems that your logic is flawed.

Re: Oracle's weird support model strikes again...

"...So in one shot, Oracle killed the hobbyist SPARC and Solaris ecosystem..."

I dont agree with this one. There are OpenSolaris forks out there, for instance OpenIndiana / Nexenta and the up coming Illumian.

Re: Oracle vs. Red Hat?

@ShelLuser

"...To me Oracle is the kind of company which knows how to charge big time for average services..."

Many people dont agree with you. The best database in the world is the Oracle DB. It has the highest performance, and best characteristics. Several world records are set by SPARC T4 and Solaris has many unique features that everybody wants to copy or port: ZFS, DTrace, Crossbow, Zones, etc. Oracle has some of the best tech, and highest performing in the world.

Oracle is not like IBM that charges astronomical amounts for low performance. For instance, the IBM Mainframe cpus are slower than a decent x86 cpu, and a Mainframe costs millions. Or, when IBM charged $500.000 for one POWER6 P570, and IBM needed six P6 servers to match one Sun T5440 at $76.000 - in Siebel v8 benchmarks. So if you want medium/low performance for a high price, you go to IBM.

What Oracle does not have, is low prices. You get what you pay for. There is a reason Oracle has recently increased the prices, whereas IBM has lowered prices. Lowering prices is a sign of weakness, if you give something away for free, you are desperate. Increasing prices are sign of strength, you are confident in your products because people will continue to buy.

Re: Re: Smart Drive

You know that raid-6 has lot of issues, right? The data might be corrupted, without the hardware/OS knowing it.

Re: @Jean-Luc: Your knowledge of C++ Applications is limited

"...C++ is used in many more ways than you point out. Think of real-time stock exchange servers and the corresponding "quant" trading applications, which need to respond in the order of 1ms...."

Well, several of the fastest stock exchange systems in the world are developed in Java. For instance the NASDAQ stock exchange is pure Java. It is among the fastest in the world, with latency of 0.1 ms and extreme through put. Java is fine if you need extreme performance.

@Wunderbar1

PS. Exa-xxx servers use Linux or Solaris. Some customers require Linux, instead of Solaris. I dont know if Oracle Linux yet supports SPARC T4, I dont think that is the case. So, Linux does not support T4 yet, so Exa-xx can not use T4.

Also, Exa-xxx servers use the beta version Solaris 11 Express. They dont yet use the sharp release Solaris 11. Why are they still using Solaris 11 Express? Maybe it takes time to test everything? If Oracle will not release Exxa with sharp Solaris 11 which is a small upgrade, why do you think Oracle will change to T4 which requires large surgery? Maybe you are just plain wrong in many of your conclusions.

Thus, SPARC will find it's way into Exa-xxx but it will take time. Let us go back to this discussion when some time has passed. You are just wrong, thinking it takes no time to upgrade. These stuff are Enterprise and that takes time. Same with Java, it is not upgraded quickly, Java is enterprise.

@Wunderbar1

Now you are dismissing all these links I provide, because they are Sun related. Do you think that is unbelievable that IBM would patent troll, just as described? Do you think it is unbelievable that IBM would patent troll on a silly patent as this?

"...The [IBM ]chief blue suit orchestrated the presentation of the seven patents IBM claimed were infringed [by Sun], the most prominent of which was IBM's notorious "fat lines" patent:To turn a thin line on a computer screen into a broad line, you go up and down an equal distance from the ends of the thin line and then connect the four points. You probably learned this technique for turning a line into a rectangle in seventh-grade geometry, and, doubtless, you believe it was devised by Euclid or some such 3,000-year-old thinker..."

Do you really think this story could not happen? IBM would never behave like this? Or... could maybe this story happened? IBM does indeed abuse their patents?

.

There are many stories about IBM abusing patents. One example, Mainframe division. IBM has sued other numerous times to sustain their monopoly on Mainframes, for instance the word "FUD" comes from IBM FUDing other companies. Without IBM, no "FUD" word:

http://en.wikipedia.org/wiki/Fear,_uncertainty_and_doubt#Definition

You know, all these stories are written by IBMs victims; company AAA, company BBB, company CCC. etc. It must be very convenient to dismiss all of these stories because "AAA got sued by IBM, what side of the story do you think he is going to take?". is this really your reason to dismiss all these horrendous stories about patent abuse? Are you serious, or are you trolling?

"Never trust the defender, because he will try to tell bad stories about the aggressor!" - or? Normally, one should listen to both sides, but you are saying "only listen to IBM side". Some would call this... slightly biased, yes?

@Wunderbar1

"...[Oracle] had no interest in their hardware..."

Maybe that was true some time ago, I dont know. You dont know either. But fact is that Oracle is today investing more in Solaris and in SPARC than Sun ever did. This is official. Larry praises Solaris as the best Unix out there. And as we have seen, IBM has done nothing new on the Operating System side for the last decades. AIX is not developed anymore - it is only maintained. IBM is betting on Linux instead and increasing R&D heavily on Linux. Oracle is targeting POWER and AIX with SPARC and Solaris. That is a fact, Larry wants to obliterate POWER and AIX. Larry can not go to war without investing heavily in SPARC and Solaris.

.

.

"...You are the worst at citing sources in the history of the world. Do you think we will just accept them at face value and not click the links?..."

This is not very clever of you. The ONLY reason to provide links, is that other check them up. That is what you do in academia, provide sources so others can check them up and so that we can have a debate. That is the reason I provide links. Why do you think it is important to provide references in academia? Hope that no researcher will check up the references??? Maybe you should go to academia first, before you draw strange conclusions? If you had gone to uni, you would have provided links and references so that all of us could read them, contemplate and raise objections in a debate and discussion. Otherwise, if you are not providing references, you are just FUDing and Trolling.

So now you are complaining that I provide links, whereas you dont provide references. At the same time, you are calling me Troll! :o) The irony. You dont really understand what you are saying. Where are your references? Nowhere. That is pure FUDing and Trolling.

Ergo, Wunderbar1 and Allison Park is FUDing and Trolling. Because I am accused of providing references - I can not be FUDing nor Trolling. :o)

@Wunderbar1

"...You are definitely a troll. Pulling a 10 year old comment made by an IBM Software Group executive that is not involved with AIX which is immediately retracted by the AIX executive is the heart of trolldome..."

Great troll I am. At least I am posting credible links, you, on the other hand, is making things up, and distorts and FUDs: "why does not ExaLogic use T4? The reason is because Oracle lies about T4 benchmarks, it is slow!". The reason ExaLogic does not use T4, is because ExaLogic has many cpus, and T4 only scales to 4 sockets. Now, stop the FUD about T4, will you? It is a well known fact that T4 is more than 2x faster than POWER7 on TPC-H.

Regarding the IBM link where several IBM people talks about killing off AIX, another executive says it is on a multi decade time frame. The first decade has passed. I dont see where I am lying or trolling? At least I am only citing IBM on stuff. That is not trolling nor FUDing.

I would like to see you and Allison Park showing credible links, just as I am doing. But you will not do that. Why? Only trolls make up FUD, and dont post trolls. And also, they are quick to yell troll on others, even when they post links directly from IBM. I mean, how much more credible can it be? It is impossible to be more credible. I am not making things up, just read all my links. Where are your links? Can you just post one single credible link? No?

@Wunderbar1

Continuing:

Or another one of IBMs result to brag over recently. IBM claimed to have simulated a cat brain and it was highlighted everywhere in IT-media. An expert on the area, found it was a scam:

http://www.wired.com/dangerroom/2009/11/darpas-simulated-cat-brain-project-a-scam-top-neuroscientist/

Have you heard about the "Worlds fastest cpu" z196?

http://www-03.ibm.com/press/us/en/pressrelease/32414.wss

"IBM (NYSE: IBM) today announced details of the world's fastest computer chip -- the microprocessor in a new version of the IBM mainframe that begins shipping to customers on Sept. 10."

But surely you did not miss that the z196 at 5.26GHz and L1+L2 cache at nearly 200MB, is slower than an high end x86. How can it be the worlds fastest cpu? How can the z196 mainframe with 24 cpus, replace 1.500 servers which use x86 cpus? Normally, a slow cpu can not replace a faster cpu - unless IBM is doing the marketing.

Another example of lies from IBM. IBM claims the POWER7 has 153GB/sec band width. It is also a lie:

http://realworldtech.com/beta/forums/index.cfm?action=detail&id=122708&threadid=122683&roomid=2

"...each POWER7 CPU in Power770/Power780 drives no more than 8 DIMMs over 8 fully-buffered channels. So simultaneous reads/writes on the same channel are not possible. 8channels*8 bytes/channel * 1.067 GT/s = 68 GB/s peak. In Power750/755 you have ~half of that. There are also published STREAM Triad scores for Power755 - 121,940 MB/s for 4 sockets = 30.5 GB/s per socket...."

Never. Ever. Trust. IBM.

@Wunderbar1

Sure, I agree that IBM did lot long time ago. Specially in virtualization. Yes, that is correct.

But recently, what cool have IBM done in Operating Systems? What new tech does AIX have, that everyone wants and drools over? Nothing. AIX only copies from others nowadays. Development has slowed on AIX, I have read in an article here. It scales bad, recently AIX had to be rewritten to be able to handle P795's few 32 cpus. IBM is betting on Linux, and increasing resources on Linux. So what cool have IBM done on AIX that everyone wants? Nothing.

OTOH, Solaris invents, and everyone copies or ports from Solaris. ZFS is unique because it protects against data corruption. DTrace is unique because it allows of observability unheard of. Zones is copied both in AIX and in Linux. etc.

Sure IBM has lot of patents. But why? Just to patent troll! That is the reason. Much of the IBM patents are worth nothing, but only used to black mail other companies for money. Here is one incident where IBM tries to black mail Sun for some ridiculous patents. Sun explains those patents will not hold in court. IBM's attournes says "ok maybe you are right, but we have 10.000 more patents, do you want us to find some patents you really breach or are you going to pay us?". Sun paid, and IBM went on to the next company on their hit list.

http://www.forbes.com/asap/2002/0624/044.html

On another occasion, IBM sued Sun for an equally silly patent on RISC, which said "if you make it simpler, it is more efficient". Sun nearly went bankrupt. After that, Sun started to patent everything, just for defense against patent trolls such as IBM:

http://nighthacks.com/roller/jag/entry/quite_the_firestorm

Or other patent trolls such as Microsoft or Apple:

http://www.osnews.com/story/22981/Schwartz_Recalls_Apple_Microsoft_Patent_Threats_Against_Sun

@Allison Park

I remarked that you never show any references. In response to my remark, you show a link to prove that in this case, you do indeed show references.

http://www.scribd.com/doc/79962880/HP-v-Oracle-Amended-Cross-Complaint

I did not really understand your point. Can you clarify it? That link only shows jibberish. What is your point, posting that link? Can you clarify the connection? What do you want to prove, and what does your link prove?

@Wunderbar1

Regarding SPARC T4 and Exalogic. There is no contradiction, it is just you drawing wrong conclusions.

One of the reasons that T4 is not in the Exa-xxx server line, is because T4 is new. It takes long time before you change an enterprise server hardware configuration.

Another reason is that T4 does only scale to 4 cpus. Exa-xxx has many more cpus than 4, so you can not use T4 in it.

The coming T5 will scale to 8 cpus. And future Niagara cpus will scale to 64 cpus, with 16.384 threads.

Thus there are no contradiction. Ok?

@Wunderbar1

Post continued here:

Regarding why Oracle is not using SPARC T4 in their Java servers. Well, the T4 is new. It can take years before they can fully support changes in such servers. It is desktop things. This is enterprise. It takes long time before any changes are made. And still, the T4 is more than twice as fast than POWER7 on TPC-H.

.

.

Regarding when IBM is going to kill AIX.

POWER6 servers where several times faster than x86 and costed 5-10x more.

POWER7 servers are ~10% faster in some benchmarks and only cost 3x more than x86.

POWER8 servers, will they be just as fast as x86? No one knows, but that is a reasonable guess if we look at historical data and analyze the trend. In that case, IBM has to lower the price even more. Maybe 1.5x. Why would IBM continue sell such low margin servers? Larry Ellison said that he does not care if x86 server sales go to zero, because it is low margin. Larry prefers high end black boxes, tailored to performance and RAS. That is where the margin is. How long do you think IBM will continue doing expensive and slower POWER cpus, when x86 has caught up? Not for long, when IBM starts to loose money.

One way forward could be if IBM ported AIX to x86, but that will not happen. Instead, IBM has officially said that AIX is going to be replaced by Linux. You better start off learning Linux, boys. Intel and AMD is pouring R&D on x86, much more than IBM can do. IBM is a profit hungry company, IBM will not keep doing POWER just for idealogical reasons. x86 will catch up. POWER will die. Why keep AIX around in that case? AIX will die, IBM has even confirmed that. Of course, this is on a multi decade timeframe. The first decade has passed. In another 10 years, x86 will be faster. And cheaper. And better RAS.

OTOH, Solaris runs on fast and cheap x86. HP-UX is not looking good. AIX is going to be killed off, IBM has confirmed this. Solaris is open sourced and will live.

@Wunderbar1

You are posting so much weird things I dont know where to start. I will select a few things, as I have work to do.

First you claim that Solaris copied from AIX. That is not true. It is true that IBM had done great innovations long time ago, for instance, in virtualization. It is true that Solaris copied LPAR and called it LDOMs. But of lately, IBM has not done anything innovative at all. Nowadays IBM is just copying others. For instance Probevue, WPAR, etc. Have you heard about the Sun BlackBox? It is a container full of servers, and IBM copied that too. IBM mocked Sun for Niagara cpus having many cores, and said that "cpus with only one strong core is best, because databases prefer strong cores". IBM talked about stronger single cores, clocked at 7-8GHz. But how does the POWER7 look like? Many lower clocked cores, just like Niagara!

What have IBM done lately, that is cool tech wise? Nothing.

.

.

Regarding if Oracle cheats with using SSD disks, etc. Well, that is not cheating. Everyone can use SSD disks but even if IBM does that, IBM can not break Oracle's records. OTOH, IBM cheats. IBM claims that one Mainframe can replace 1.500 x86 servers. Well, it turns out that x86 servers idle and the Mainframe is 100% loaded! That is cheating.

Also, the z196 Mainframe cpu, dubbed "worlds fastest" is actually quite slow. It is much slower than a high end x86 cpu. So what happens if some of the x86 servers do some work? Then the mainframe chokes, and certainly it can not replace 1.500 x86 servers doing work. If I boot 5 mainframes that idles, on my laptop, can I claim that my laptop can replace 5 mainframes? No, that would be FUD, right? You will never see SPECINT benchmarks for Mainframe cpus, from IBM - for a reason. They are slow cpus. In fact, if you emulate a Mainframe on a x86 server, you can get 3.200 MIPS - which is a decent Mainframe. Thus, a few slow 196 cpus can not replace 1.500 faster x86 cpus.

@Allison Park

Are you calling ME a troll? You are too funny. At least I only cite sources I can refer back too, and never make anything up myself. Everything I say, I can backup. That is not trolling, I am only citing others. You are shooting the messenger, as I am only the messenger.

If you mean that IBM is trolling in the article I posted, fine. But dont call me Troll. You on the other hand, is a well known troll and FUDer. You claim false and negative things, only to put IBM's competitors in bad light. And when someone ask you for references, you can never show any. The reason you never show any references, is because you make everything up. If you read the wikipedia article on FUD, you will see that your behavior fulfills the definition of FUD.

Me on the other hand, is never FUDing, because I always back up my claims with trustworthy links, such as links from IBM. If you read the wikipedia article, this is not FUD. In fact, it is the opposite; sane and sound argumentation.

Phil 4

And please dont forget that IBM is going to kill AIX too:

http://www.zdnet.co.uk/news/application-development/2003/01/29/ibm-linux-will-replace-aix-2129537/

x86 is today only 10% slower than POWER7 on some benchmarks (for instance, SAP). When x86 catches up and surpasses POWER, why would you buy expensive and slower POWER gear, than going with faster and cheaper x86? Then people will stop buying POWER, and sales will dwindle.

And because IBM only does high margin business, IBM does not want to lower POWER prices even more, because then POWER will be low margin business. And when that happens, POWER will be killed. Together with AIX. Just so people know this and be careful when examining POWER alternatives.

@A.C

"...Care to actually reference any sources (and I don't meen uncorroborated quotes with no context on Oracle's website)?..."

Allison Park has never referenced any sources. I have asked numerous times. According to wikipedia; making unsubstantiated negative claims is FUD.

Sun X4500 Thumper

I think it was Sun that first sold 4U storage servers with disks "standing" up. Normally, disks are inserted into the front, but Sun storage servers are lowered down from above, by removing the top lid. Therefore x4500 could have 48 disks in 4U.

Normal 4U storage chassis, such as the Norco 4224 disk chassi which is 4U, only holds 24 disks. Inserted into the front.

Later, other vendors copied Suns design. Backblaze holds 45 disk in 4U. There is another vendor that holds 60 disks in 4U chassis.

.

.

Regarding this 10 Exabyte, there are several problems. First, the data will rot, and there will be lot of data corruption. RAM sticks have data corruption caused by cosmic radiation, current spikes, etc - therefore ECC RAM is necessary to detect flipped bits and correct them. The same bit rot will occur on all these disks, so you need checksums to detect and correct randomly flipped bits.

Second, 10 Exabyte is too much for most filesystems, such as BtrFS, because they are 64 bits. You need 128 bit filesystems to handle 10 EB in one single namespace.

Incidentally, ZFS solves both of these problems above.

@gerdesj

"...Finally, no matter how fancy it is, you'd better back it up properly and give it a UPS and a good hardware RAID or SAN if you care about your data...."

Backups are good, but how do you know that the backup does not contain corrupted data? I have heard of old backupped files, and long time after, the files are not readable. The zip file is not extractable, the media mp3 is not playable, etc.

A backup needs checksums to detect and protect against these data corruption errors. That is the reason MS is adding checksums for metadata. However, that does not suffice, the data need also to be checksummed. Thus, ReFS is not that safe. As research shows, NTFS is not safe either. Nor is XFS, ReiserFS, JFS, etc which researchers prove. There is a reason ECC RAM uses checksums, so should filesystems.

Thus, a backup on a hardware raid or SAN does not suffice, and you might still have corrupted data. There are many texts on this, I hope you have not missed them.

@A.C

"...The hardware ECC in every drive, however, is [protecting against data corruption]..."

Well, that is not true. If you look at the spec sheet of any Enterprise disk, for instance Fibre Channel disks, or SAS disks you will see typically

"1 irrecoverable error in every 10^16 bits read"

This proves that ECC on disks does not protect against bit rot. The same thing with hardware raid - they do not protect against data corruption. There are no checksums for that.