Farewell To OpenSolaris. Oracle Just Killed It Off.

[LightningCrash]

There are 1024 socket machines and even 2.4 kernel supported 256 sockets within a single system image.

And no, the SSI is a layer that runs on top of an existing Linux kernel on a smaller machine.
So there are not even any 1024 socket machines. The Altix 4700 is a networked collection of dual socket blades.
If your criteria excludes multiple threads per CPU as not being real CPUs it should also exclude SSI, which is just a networked series of smaller individual SMP systems.

[LightningCrash]

I was just referring to heise-online like I said before. I suppose it should be documented somewhere which is the upper limit of CPUs Solaris can scale to.

Oh, I'm aware of this. It's Kebabbert who's claiming Linux scales bad when comes to vertical scaling. He's also claiming many other, strange and "trolly" things, but you've probably missed all of these. It seems you're sane, so take look at this, please:

http://www.redhat.com/f/pdf/SAS_2010_Benchmark_06-15-2010.pdf

It shows RHEL 5.4 scales linearly up to 64 cores, so who's the person who says Linux doesn't scale or scales very bad? While it scales linearly and he says "no it doesn't scale" or "it "scales badly" etc. then who's he?

I don't 'poobah' when someone says sane things. Like I mentioned I was according to heise-online before. You should realize Kebabbert's trolling and what he says is really not relevant in any way. :> Btw. RHEL 6 supports up to 64,000 cores in a single system image and 128 TB of memory for the kernel and 128 TB for the userspace while Solaris 11 will support 64TB (according to h-online). Just a note.

Lol at trying to drag me into the vertical scaling.

SSI != hardware
cores != sockets

What's the largest number of sockets you can get in one single Linux machine? SSI doesn't count as CPUs, for the same reason that the T2's threading doesn't count.

[kraftman]

Lol at trying to drag me into the vertical scaling.

Nope, it was just the original reason of the 'conflict'. Kebabbert's claiming Linux scales badly when comes to vertical scaling.

SSI != hardware
cores != sockets

What's the largest number of sockets you can get in one single Linux machine? SSI doesn't count as CPUs, for the same reason that the T2's threading doesn't count.

I'll tell you when I found. I've got numbers for RHEL5 only, but it's using very old kernel.

[kebabbert]

Oh, I'm aware of this. It's Kebabbert who's claiming Linux scales bad when comes to vertical scaling.

I am only quoting three Linux scaling experts that say:
1) There are lots of FUD from Unix vendors that Linux scales bad
2) Linux scales very good
3) Google uses 10.000 Linux computers which proves Linux scales good (this is horizontal scaling)
4) Linux also scales very good vertically. "Linux v2.6 will even scale to 16 cores!!!!"
5) That Linux scales bad is just FUD.
The Linux scaling experts from RedHat, etc - dispels the Linux scaling FUD:

http://searchenterpriselinux.techtarget.com/news/929755/Experts-Forget-the-FUD-about-Linux-scalability

It is only Linux scaling experts that mix up vertical scaling and horizontal scaling. Nobody denies Linux scales good horizontally (just add another PC to a network, just like Google does). The Linux problem is vertical scaling, some work loads are very easy to do in parallell (just add another PC, like the SETI project), and others are very hard to do in parallell (P-complete problems).

If the Linux scaling experts think that 16 core vertical scaling is very good - that says all. Only linux people thinks that 16 core scaling is good. Hilarious.

He's also claiming many other, strange and "trolly" things, but you've probably missed all of these. It seems you're sane, so take look at this, please:

http://www.redhat.com/f/pdf/SAS_2010_Benchmark_06-15-2010.pdf

It shows RHEL 5.4 scales linearly up to 64 cores, so who's the person who says Linux doesn't scale or scales very bad? While it scales linearly and he says "no it doesn't scale" or "it "scales badly" etc. then who's he?

Great. Have you read your SAS benchmark you link to?

The official SAP benchmark I talked about earlier, when Linux got 87% cpu utilization, and Solaris got 99% - Solaris won. Linux used faster hardware (faster cpus, faster RAM sticks) and still lost. That is because the CPU utilization was so low with Linux machine. Solaris had higher CPU utilization and won on slower hardware.

Have you read your SAS benchmark? The CPU utilization is very low. Have you seen the graphs of the cpu utilization? The graph is never maxed. Most of the time, it is below 40% or so. It reaches as high as 95% in, like, one short spike or so. In SAP, Solaris had 99% cpu utilization, of course Linux lost. Here, CPU utilization is below 30%, quite far from 99% or even 87%.

I don't 'poobah' when someone says sane things.

What a joke. Do you want me to quote your weirdest things? I can do it. Just ask me someone.

For instance, I linked to SAP official benchmarks, and you rejected them as "FUD from Sun, and Propaganda". You also said that SAP has partnership with Sun, and therefore Linux is punished. I showed that SAP has partnership with several Linux companies, why is Solaris not punished? Hilarious.

Or when I link to research papers on data corruption - and you deny I linked to research papers. I posted the links again, but you denied they exist. You saw the links, but you said that "kebabbert lies and FUD, there are no research papers" - but all who tried all links saw the research papers.

So you dont 'poobah' when someone says sane things? Like, when you claim that Solaris scales only to 64cpus, but several others have explained to you that it is not true, you still continue to FUD about Solaris scalability. You have no idea about Solaris scalability. Why did you use the number 64? Why not a number much lower? Like... 16 cpus? I suggest you tell everyone that Solaris only scales to 16 cpus. You are already FUDing, so you can as well FUD seriously, say that Solaris scales to 16 cpus.

You confessed that you FUD. I can quote when you say so.

Like I mentioned I was according to heise-online before. You should realize Kebabbert's trolling and what he says is really not relevant in any way. :> Btw. RHEL 6 supports up to 64,000 cores in a single system image and 128 TB of memory for the kernel and 128 TB for the userspace while Solaris 11 will support 64TB (according to h-online). Just a note.

FUD again. Oracle will release a machine with 64TB RAM. It says nothing about how much RAM solaris will support. When IBM released the first IBM PC with 64KB RAM, it does not mean that MS-DOS only can support 64KB RAM. Wrong again.

On the other hand, I have told you that if you compile Linux to a computer with 1024 cpus (which does not exist) it does not mean that Linux can use all CPUs. With all the limitations and bad scaling in Linux, I doubt you can use a big Linux machine well. Just talk to the sysadmins, I have done it. They all say that Linux sucks, when comparing to big Enterprise Unixes. I do not lie on this, just talk yourself to Unix enterprise sysadmins.

[kraftman]

I am only quoting three Linux scaling experts that say:
1) There are lots of FUD from Unix vendors that Linux scales bad
2) Linux scales very good
3) Google uses 10.000 Linux computers which proves Linux scales good (this is horizontal scaling)
4) Linux also scales very good vertically. "Linux v2.6 will even scale to 16 cores!!!!"
5) That Linux scales bad is just FUD.
The Linux scaling experts from RedHat, etc - dispels the Linux scaling FUD:

http://searchenterpriselinux.techtarget.com/news/929755/Experts-Forget-the-FUD-about-Linux-scalability

It is only Linux scaling experts that mix up vertical scaling and horizontal scaling. Nobody denies Linux scales good horizontally (just add another PC to a network, just like Google does). The Linux problem is vertical scaling, some work loads are very easy to do in parallell (just add another PC, like the SETI project), and others are very hard to do in parallell (P-complete problems).

Saying Linux doesn't scale good vertically (at least good, but while it scales linearly or when it riches above 90% CPU utilization it scales great) is FUD. When comes to point four, it's an great example of your idiocy and FUD. They're talking about the first 2.6 which is about eight years older then newest one and you're ignoring this fact all the time.

If the Linux scaling experts think that 16 core vertical scaling is very good - that says all. Only linux people thinks that 16 core scaling is good. Hilarious.

Like I said it's your idiocy. The talk was about eight years old kernel.

Great. Have you read your SAS benchmark you link to?

The official SAP benchmark I talked about earlier, when Linux got 87% cpu utilization, and Solaris got 99% - Solaris won. Linux used faster hardware (faster cpus, faster RAM sticks) and still lost. That is because the CPU utilization was so low with Linux machine. Solaris had higher CPU utilization and won on slower hardware.

Yes I had, but you're ignoring the fact this only shows Solaris server has won over Linux server (in theory of course, because if you compare the price maybe Linux server offers greater performance then Solaris server at lower costs). The fact is this Solaris server is much, much more expensive then Linux server, so they're aiming at different markets (these servers, not operating systems). I suppose Solaris server was especially designed to scale great on this machine, it has also twice amount of memory then Linux server.

Have you read your SAS benchmark? The CPU utilization is very low. Have you seen the graphs of the cpu utilization? The graph is never maxed. Most of the time, it is below 40% or so. It reaches as high as 95% in, like, one short spike or so. In SAP, Solaris had 99% cpu utilization, of course Linux lost. Here, CPU utilization is below 30%, quite far from 99% or even 87%.

Here you're showing your stupidity. Take a look at your CPU usage right now. If you're not doing expensive tasks then your CPU is not utilized in 100%. This test I gave shows Linux server scales linearly up to 64cores (maximum number of cores on this server). Utilization reached 95% when it was doing the most demanding tasks. You didn't read carefully, there's explanation why utilization wasn't even higher:

The I/O output from 8-Socket server exceeded the I/O capacity of test storage configuration. Server utilization would have been significantly greater with more storage controllers.

It's probable Linux server from the SAP had less storage controllers then Solaris server, so it didn't rich such utilization, but it was cheaper.

What a joke. Do you want me to quote your weirdest things? I can do it. Just ask me someone.

For instance, I linked to SAP official benchmarks, and you rejected them as "FUD from Sun, and Propaganda". You also said that SAP has partnership with Sun, and therefore Linux is punished. I showed that SAP has partnership with several Linux companies, why is Solaris not punished? Hilarious.

I showed you SAP attacking Linux, so I think it's fair to have some objections.

Or when I link to research papers on data corruption - and you deny I linked to research papers. I posted the links again, but you denied they exist. You saw the links, but you said that "kebabbert lies and FUD, there are no research papers" - but all who tried all links saw the research papers.

I was denying such papers backup you're claims. Of course, I could missed some papers, because your posts were very chaotic. When I was saying you lie and FUD it's not because (or not only) I was saying you didn't show the papers.

So you dont 'poobah' when someone says sane things? Like, when you claim that Solaris scales only to 64cpus, but several others have explained to you that it is not true, you still continue to FUD about Solaris scalability.

I have no problems with this now. Like I said I was according to heise-online. It's a matter if I believe heise-online or some people here. I usually don't believe you, but I have believed in what LightningCrash was saying, because he seems to be sane. You still continue the FUD about Linux vertical scalability. Even, if you saw it reaches 95% of CPU utilization on bigger machine then the one from SAP (and while it's explained why Linux server didn't reach even better utilization) you still spread the FUD.

You have no idea about Solaris scalability. Why did you use the number 64? Why not a number much lower? Like... 16 cpus? I suggest you tell everyone that Solaris only scales to 16 cpus. You are already FUDing, so you can as well FUD seriously, say that Solaris scales to 16 cpus.

You have no idea about Linux scalability. I used number 64, because heise-online used this number too. So, this what you're writing next is just a bull and lie.

You confessed that you FUD. I can quote when you say so.

I don't FUD intentionally like you do. It's not my fault somebody claims current Solaris servers support only 64CPUs. :>

FUD again. Oracle will release a machine with 64TB RAM. It says nothing about how much RAM solaris will support. When IBM released the first IBM PC with 64KB RAM, it does not mean that MS-DOS only can support 64KB RAM. Wrong again.

When there are two unrelated SAP benchmarks which says nothing about what maximum CPU utilization Linux can reach on big machines you're using this to FUD about Linux. When IBM makes the test when they benchmark some scenario (like handling some number number of clients) and Linux does the job using only 50% of CPU utilization it does not mean Linux can reach only 50% of CPU utilization. FUD again.

On the other hand, I have told you that if you compile Linux to a computer with 1024 cpus (which does not exist) it does not mean that Linux can use all CPUs. With all the limitations and bad scaling in Linux, I doubt you can use a big Linux machine well. Just talk to the sysadmins, I have done it. They all say that Linux sucks, when comparing to big Enterprise Unixes. I do not lie on this, just talk yourself to Unix enterprise sysadmins.

You don't have to compile Linux to support 1024CPUs. It can handle 4096CPUs without providing significant overhead when used on a much smaller machines (like two cores etc.). You FUD again saying about limitations and bad scaling. I heard it's Solaris which sucks, it cannot take advantage of using RCU and I heard Linux scales crap out of Solaris, so I doubt Solaris can scale as good as Linux on the biggest machines.

[LightningCrash]

I heard it's Solaris which sucks, it cannot take advantage of using RCU and I heard Linux scales crap out of Solaris, so I doubt Solaris can scale as good as Linux on the biggest machines.

As you said yourself, it's just simple hearsay.

[jadrevenge]

As you said yourself, it's just simple hearsay.

Is it just me, or does this seem to be headed for "My mum is better than your mum" or "Your mum is a slag!" kinda dialog.

[LightningCrash]

Is it just me, or does this seem to be headed for "My mum is better than your mum" or "Your mum is a slag!" kinda dialog.

I think it's been there for a while.

[kebabbert]

Saying Linux doesn't scale good vertically (at least good, but while it scales linearly or when it riches above 90% CPU utilization it scales great) is FUD. When comes to point four, it's an great example of your idiocy and FUD. ...They're talking about the first 2.6 which is about eight years older then newest one and you're ignoring this fact all the time.

So what? The Linux scaling EXPERTS say that 16 cores is good scaling. In the interview, the EXPERTS try to dispel all the Unix vendors FUD that Linux scales bad. They say "Linux scales good, contrary to all the FUD, in v2.6 Linux will scale to 16 cores!". What does that prove to you?
-Hey! Dont come here lying about Linus scaling bad! It scales excellent, in v2.6 it will scale to 16 cores! You hear me? SIXTEEN CORES! Now THAT rocks! Gosh! Ok, stop telling everyone that Linux scales bad, because it is only FUD! Google uses Linux on their 10.000 pc network cluster. Linux scales excellent! You are a FUDer! Shut up!

In one word: Ridiculous. It doesnt matter if this discussion is eight years old, it only shows the Linux SCALING EXPERT's attitude. They claim Linux scales to 10.000 computers, and at the same time they say Linux scales to 16 cores - which is excellent scaling!!! Of course Linux scales better today. It may even scale to... 32 cores? I dont know. When Linux scaling experts today say that Linux scales excellent, they probably mean it scales to 32 cores today. But the step from 16 cores to 4096 cpus, is... quite a big step and not very likely? IBM highend Unix AIX does not scale well even today, after decades. It takes decades to scale well. And the developers need to have access to big machines. Linux developers mostly use 1-2 cpus with quad cores. I really do think Linux are fastest on such desktops, faster than Solaris. But they typically dont have access to large machines, and can not tune Linux to such machines.

Yes I had, but you're ignoring the fact this only shows Solaris server has won over Linux server (in theory of course, because if you compare the price maybe Linux server offers greater performance then Solaris server at lower costs). The fact is this Solaris server is much, much more expensive then Linux server, so they're aiming at different markets (these servers, not operating systems). I suppose Solaris server was especially designed to scale great on this machine, it has also twice amount of memory then Linux server.

You know what kind of servers are expensive? The servers with high RAS are expensive. You pay for RAS. Not performance. Performance is not expensive. IBM Mainframes are dog slow cpu wise, but have very good RAS. That is the reason IBM Mainframes are shitload expensive. Wrong again.

And, Linux machine used faster RAM sticks and faster CPU. The reason Linux machine did not use 256GB RAM, but instead went for the lower 128GB RAM, is because if Linux went for 256GB, then Linux must use much slower RAM sticks. I have written this many times. But you just ignore it. SAP benchmark does not require 256GB RAM, because then Linux would go for 256GB RAM as well.

If you wish, I can mail SAP technical support and ask how many GB RAM is required to complete SAP benchmark. Maybe it is 8GB RAM? Or even 12GB? Or maybe it is better that you do it. Because when I show you the answer from SAP, you will just say "lies, it is FUD, it is Sun propaganda".

Here you're showing your stupidity. Take a look at your CPU usage right now. If you're not doing expensive tasks then your CPU is not utilized in 100%. This test I gave shows Linux server scales linearly up to 64cores (maximum number of cores on this server). Utilization reached 95% when it was doing the most demanding tasks. You didn't read carefully, there's explanation why utilization wasn't even higher:

The SAS benchmark simulates many users that log on and do some simple tasks. This work load is easy to parallellize. Just like SETI client. And the work load has very low CPU utilization. This is a workload which is not taxing nor demanding. Any OS could handle this work load. For instance, let 5 independent users be assigned to one core. When there are 10 users, assing two cores. etc. Does this work load sound challenging to you? Does it sound like hard to parallellize? Do you think SETI is hard to parallellize?

We had this discussion before and llama wrote about scalability and work loads that are hard to scale (SAS benchmark is not hard to scale). You was in this thread, but still you ignore it.

http://phoronix.com/forums/showpost.php?p=62703&postcount=80

"We're not talking about clusters. We're talking about single system-image big iron, where _one_ kernel runs on a single machine with > 16 CPUs in a cache-coherent shared-memory system. The most cost-effective machines for cluster-building, in CPU power per dollar, are dual-socket quad core Intel Core2-based machines. i.e. 8 cores per node. That's great if you have a workload that has some coarse-grained parallelism, or is embarrassingly parallel, e.g. processing 100 separate data sets with single-thread processes that don't depend on each other. That's not so great if you have a lot of processes that need fine-grained access to the same shared resource. The canonical example here is a database server handling a database with a significant amount of write accesses. Otherwise you could just replicate it to a big cluster and spread the read load around. Locking for write access in a big cluster, even with low latency interconnects like infiniband, is still _way_ higher overhead than you'd get in e.g. a 4 or 8 socket quad-core machine. Even NUMA big iron is better suited for this than a cluster.

CLUSTERS DON'T COUNT AS BIG IRON. They're just a pile of normal machines. They do have their uses, though."



And you also ignore this:

http://phoronix.com/forums/showpost.php?p=85532&postcount=54

I was denying such papers backup you're claims. Of course, I could missed some papers, because your posts were very chaotic. When I was saying you lie and FUD it's not because (or not only) I was saying you didn't show the papers.

I showed my post with all research papers several times to you. And each time you denied they exists, that I FUD and lie. And you ask me to post research papers, otherwise I FUD and lie. And I post them, and you still say I FUD and lie:

http://phoronix.com/forums/showpost.php?p=140667&postcount=149

But I said the paper was there all the time.

http://phoronix.com/forums/showpost.php?p=141010&postcount=163

In short I have proved you FUD and lie about me. The paper and the rest of the research papers are there. You just deny them. Obviously, you are trolling, Kraftman. Just as usual. You also confessed you FUD, Kraftman, in a post. Here I show you lie and FUD about me, and about Solaris. Earlier you confess you FUD. this is fun. How much longer will you claim that it is kebabbert that FUDs? I have showed links and papers and research papers, and links to Linux kernel developers. There is no lies there, nor FUD. You are quite stubborn. Dont understand when you have lost a battle. Sure, continue to say that Solaris scales to 64 cpus, but that is not true. Continue to say that I did not post research papers, that is also not true. etc.

You don't have to compile Linux to support 1024CPUs. It can handle 4096CPUs without providing significant overhead when used on a much smaller machines (like two cores etc.). You FUD again saying about limitations and bad scaling. I heard it's Solaris which sucks, it cannot take advantage of using RCU and I heard Linux scales crap out of Solaris, so I doubt Solaris can scale as good as Linux on the biggest machines.

Of course Solaris does not scale. You have explained to us that 128 cpu Solaris machines such as this:

http://news.cnet.com/Fujitsu-to-debut-128-CPU-server/2100-1001_3-269806.html

does not exist. That we lie and FUD. And also, the Solaris machine with 144 cpus also doesnt exist. That is also FUD and lies. We are just a bunch of FUDers and liars everyone of us.

I think you should say that Solaris scales to 16 cpus, that is a much lower than 64 cpus. Why do you say Solaris scale to 64 cpus? If you are going to FUD about Solaris, at least FUD properly. Say Solaris scales to 16 cpus instead of 64 cpus.

[nanonyme]

Are you people still continuing this? o.O