Announcement

Collapse
No announcement yet.

Netbook Performance: Ubuntu vs. OpenSolaris

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • #41
    Originally posted by kebabbert View Post
    You dont understand what I am trying to say. I am not saying that Linux is 250 times slower than Solaris. I am not trying to say that Linux page fault is 250 times slower than Solaris. You dont understand.
    You're according to my comment about pdf (where - is and this means, 250 times faster was a joke there :>), but I already answered this before.

    I am trying to say that just recently, Linux had some great scalability problems on 64 cpu machines. Linux is still inmature on Big Iron. I bet there are lots of other areas where Linux still has scalability problems. Solaris has been doing this on Big Iron for decades. I doubt Solaris performance can be 250 times increased.
    If you read this pdf and looked at those tests 2.6.8 scales very, very well. I'm trying to say those tests you saw was fake - gnu malloc, improper kernel config etc. I now wonder what you mean. You said you don't consider Linux is 250 times slower then Solaris, but you're saying now you doubt Solaris performance can be 250 times increased. This is like Linux performance was increased 250 times, but only page fault is faster and we don't if this have big impacts on performance/scalability. Again, 2.6.8 (it will be my favorite now ).

    This is not true, but I am trying to say something like: If it is impossible to improve Solaris performance 250 times, but you can still improve Linux 250 times - which is most mature? I bet there are still large optimizations to be done on Linux. Linux is not mature, it has not scaled on Big Iron for decades yet.
    Answer Rip-Rip question please.

    About profffs. It was a joke to show this discussion is idiotic... You see - profffs. I make grammar mistakes, but not like this The problem is there's too much FUD about Linux scaling. It's even mentioned in one of the articles you gave. Sorry I ignored you're reply before. I should response to you...
    Last edited by kraftman; 08-02-2009, 05:39 PM.

    Comment


    • #42
      Originally posted by kraftman View Post
      You're according to my comment about pdf (where - is and this means, 250 times faster was a joke there :>), but I already answered this before.

      If you read this pdf and looked at those tests 2.6.8 scales very, very well. I'm trying to say those tests you saw was fake - gnu malloc, improper kernel config etc. I now wonder what you mean. You said you don't consider Linux is 250 times slower then Solaris, but you're saying now you doubt Solaris performance can be 250 times increased. This is like Linux performance was increased 250 times, but only page fault is faster and we don't if this have big impacts on performance/scalability. Again, 2.6.8 (it will be my favorite now ).

      Answer Rip-Rip question please.

      About profffs. It was a joke to show this discussion is idiotic... You see - profffs. I make grammar mistakes, but not like this The problem is there's too much FUD about Linux scaling. It's even mentioned in one of the articles you gave. Sorry I ignored you're reply before. I should response to you...
      2.6.8 kernel is NOT supported by RedHat or SUSE. Customers will not run kernels that are not supported. Period. End of sentence. It could be 1000 times better than the installed kernel and it would still not matter.

      If 2.6.8 kernel is so wonderful, then why is it not available on a commercially-supported distribution?

      Here is a typical Linux FAIL scenario:

      Customer has hard limit for transaction completion, say 5 seconds with 10000 users. Linux may complete 99.9% of transactions in 1.5 seconds, but 0.01% come back in 6 seconds, FAILING the test.

      On same hardware, Solaris completes 100% of transactions within 5 seconds and PASSES the test. Even if the average response time is twice Linux, Solaris WINS and Linux LOSES.

      Comment


      • #43
        Hard performance requirements

        In case you were wondering, here are customer scenarios that require hard transaction time limits:

        - Customer is automobile manufacturer, and cannot afford to stop assembly line to wait for query results

        - Customer is stock trader and has upstream performance promises to their customers.

        - Customer is bioengineering firm and cannot wait for samples to spoil before receiving test results.

        - Customer is Turnpike Authority and must process SpeedPass accounts to avoid traffic jams at toll booth.

        - Customer is Lottery Commission and cannot tolerate hundreds of thousands of unsatisfied customers 10 minutes before drawing.

        - Customer is package shipping company that cannot stop conveyor belts to wait for package routing information.

        - Customer is commercial airline and cannot hold flights to wait for passenger manifest generation.

        - Customer is NOT junior in long underwear playing with daddy's computer.

        Comment


        • #44
          Originally posted by frantaylor View Post
          2.6.8 kernel is NOT supported by RedHat or SUSE. Customers will not run kernels that are not supported. Period. End of sentence. It could be 1000 times better than the installed kernel and it would still not matter.

          If 2.6.8 kernel is so wonderful, then why is it not available on a commercially-supported distribution?

          Here is a typical Linux FAIL scenario:

          Customer has hard limit for transaction completion, say 5 seconds with 10000 users. Linux may complete 99.9% of transactions in 1.5 seconds, but 0.01% come back in 6 seconds, FAILING the test.

          On same hardware, Solaris completes 100% of transactions within 5 seconds and PASSES the test. Even if the average response time is twice Linux, Solaris WINS and Linux LOSES.

          You're so f*cking troll XD. 2.6.8 is very old kernel. You're a moron XD Point was this if such old kernel scales wonderfully at least up to 64 CPUs on "Big Iron" machine, newer kernels scales even better. You're talk about transactions is bullshit and it's opposite to what you said before. Thanks a lot. I wouldn't be surprised if Yotambien is hidden under those 'new registered' nicks. Similar style XD I have no doubt, 2.6.8 is my favorite one XD

          Comment


          • #45
            Originally posted by frantaylor View Post
            Here is a typical Linux FAIL scenario:

            [..trolling..]

            Solaris WINS and Linux LOSES.
            What are you talking about? Really, if I should say the same thing replacing Solaris with Linux and the other way around in that sentence, would it make *any* sense?

            You're just making something up and then you say Linux sucks? What the heck?

            Look:
            Here is a typical Solaris FAIL scenario:

            Customer has hard limit for transaction completion, say 5 seconds with 10000 users. Solaris may complete 99.9% of transactions in 1.5 seconds, but 0.01% come back in 6 seconds, FAILING the test.

            On same hardware, Linux completes 100% of transactions within 5 seconds and PASSES the test. Even if the average response time is twice Solaris, Linux WINS and Solaris LOSES.

            Comment


            • #46
              @MartjeB

              I think they/he is already full. No more feeding from my side .

              Comment


              • #47
                Rip-Rip
                "I didn't read the whole topic so maybe you've already answer to my questions. Did you compare, I mean on same hardware, with real charts, Solaris against Linux on those "Big Iron"? Also, how can you be sure that Solaris can not be improved?"

                I am not saying that Solaris can not be improved. I am saying that I doubt Solaris can be improved 250 times, just as Linux. Maybe Solaris can be improved 100% or so. But I doubt Solaris can be improved 25000% just as Linux.

                I am trying to say that Linux can be improved much 250 times, while Solaris can not maybe 2 times. I have no proof of this, but I suspect this. Because Solaris has run on Big Iron for decades, whereas Linux has not. During these decades, Solaris has corrected all those obvious slow downs, that Linux suffers from.

                As I said, I have no proof of this. But I suspect this. And it sounds reasonable, that a Operating System that has run for decades on Big Iron has ironed out the most obvious slow downs, and now only smaller optimizations is left to do? While a fresh Operating system still has lot of large improvements to do? Is it reasonable?

                The problem is that not all Linux kernel devs have access to Big Iron, only a few has. Whereas all Solaris kernel devs have always access to Big Iron. The majority of Linux patches will not be about Big Iron, only about 2-4 CPUs and clusters.


                Kraftman,
                Regarding your pdf showing that "Linux scales very well" - fine. But I doubt that Linux scales very well in general. It seems that Linux scales well on one benchmark. Linux DOES scale well, on large clusters. That is well known fact. Everyone agrees on this. But on Big Iron, Linux sucks. Here are some Linux experts on scalability. They say Linux sucks on big Iron, but is very good on clusters:
                http://searchenterpriselinux.techtar...929755,00.html
                You hear that Linux is good on 10.000 cpus, no Big Iron has 10.000 cpus. They have only 100 or so CPUs. If they talk about scaling on 10.000 cpus, they always mean clusters, which basically is a network.

                Comment


                • #48
                  Okay, my point is subtle so please read carefully. I do not dispute that Linux scales well. I do not dispute that Linux will handle a workload faster than Solaris.

                  What I am saying is that with Linux, all threads do not receive equal CPU access. Some get more than others. There is a lot of variation. I can see this myself on my 8-core box. I can start up 8 "tar czf" commands on a big directory and watch them all run. They do not all get the same amount of CPU. They do not all finish at the same time. Some of them take much longer. Solaris is less efficient, but the kernel is better at making sure that all the threads are getting CPU equally.

                  No you do not know me. I am not who you think I am.

                  I have been writing code for Unix since BSD 4.2 on a PDP-11. I still have an SLS CD-ROM with 0.99pl3 kernel that I ran on my '486. I have used many kinds of UNIX boxes from Data General to SGI workstations to huge AIX machines. I love Linux, it runs on all my computers. I run Fedora on my machines and I submit bug reports and patches.

                  All I am saying is that Solaris is better if you need your CPU threads to be distributed evenly. Otherwise it sucks. I don't want it on my desktop and I would not run a web server on it. But I will run a big SQL database on it if my users need their SQL queries to complete in a timely manner every time.

                  One thing that does bug me about Linux is the regressions. I go to a lot of trouble to set up a box in a certain way. Then I run a software update and now something does not work. For instance my laptop used to run Google Earth just fine. I did a software update and now it does not work. Once I could not even log in after a software update because of a change to a gnome config file, it took far too much of my time to figure that out.


                  I am a busy guy and I don't have time to dig into this kind of stuff, I have enough problems. My employer supports 20 platforms and I need to work with test machines for every one. You are very lucky that you do not have to deal with VMS, it is a nightmare.

                  There is also a lot of unfinished stuff in Linux. I have to work with IPv6 and the kernel support is perfect. But userland has not caught up. Stuff that just works in IPv4 needs to be tweaked for IPv6. The DHCPv6 client sometimes gets stuck and I have to kill it. I also had to patch SELinux to get the DHCPv6 client to work. I don't have these problems with IPv6 in Solaris or even Windows, everything works great as is.

                  Can we please put a lid on the swearing and name calling?

                  I am here because I want to learn how to make my Linux machines run better. I thrash on them hard for my job and I spend too much time waiting for them. I have learned a lot over the years about performance and sysadmin and I would like very much to share. Unfortunately it seems that this site has many spiteful people who get their kicks by insulting and swearing and name-calling. I don't have time for this either.

                  Comment


                  • #49
                    frantaylor,
                    Interesting post. You state that Solaris is slower than Linux from your experience. That is interesting.

                    I wonder, have you tried both OS on large machines with lots of CPUs and large load? From reading articles I get the impression that Linux is good on few CPUs but jerks and stutters when having to use many CPUs, whereas Solaris runs evenly on many CPUs. Maybe Solaris flows slower than Linux on few CPUs, but I suspect Solaris flows better on many many CPUs, it doesnt hickup. Whereas Linux runs fast on few CPUs and hickups and stutters on many CPUs, it doesnt flow smooth and even.

                    Linux is like a small car, it is fast but as soon it has to climb a height or carry large load, the car looses speed. Solaris is like a big loader, it maybe is slower on light load but doesnt slow down when climbing big heights or carrying large loads. The big loader with the strong motor just runs and runs, not matter what work to do.

                    Kind of. What does your experience say about this analogy?

                    Comment


                    • #50
                      Originally posted by kebabbert View Post
                      frantaylor,
                      Interesting post. You state that Solaris is slower than Linux from your experience. That is interesting.

                      I wonder, have you tried both OS on large machines with lots of CPUs and large load? From reading articles I get the impression that Linux is good on few CPUs but jerks and stutters when having to use many CPUs, whereas Solaris runs evenly on many CPUs. Maybe Solaris flows slower than Linux on few CPUs, but I suspect Solaris flows better on many many CPUs, it doesnt hickup. Whereas Linux runs fast on few CPUs and hickups and stutters on many CPUs, it doesnt flow smooth and even.

                      Linux is like a small car, it is fast but as soon it has to climb a height or carry large load, the car looses speed. Solaris is like a big loader, it maybe is slower on light load but doesnt slow down when climbing big heights or carrying large loads. The big loader with the strong motor just runs and runs, not matter what work to do.

                      Kind of. What does your experience say about this analogy?
                      That sounds right to me. Although even with all of that stuttering, Linux does manage to give more of the CPU to the threads and consume less in the kernel. It's just that it's not even. Some threads get lots of CPU and some get very little. It's the threads that are starved of CPU that are the problem. Like I said, it doesn't matter if Linux gets the job done faster on average. What matters is that ALL the threads finish in a timely manner.

                      Thanks for your response. There are lot of people on this board who take any Linux criticism way too personally. For goodness sake, it's just software. I get the feeling that most of these foul-mouthed people are just bored trust-fund babies with too much money to spend on their watercooled gaming rigs.

                      And who knows, maybe someday Linus will start paying attention to people who know something about process scheduling. There's always hope.

                      Comment

                      Working...
                      X