Slashdot Mirror


Solaris' Dtrace in Detail

paulkoan writes "The Register has a further details about the new Dtrace systems utility bundled with Solaris 10, along with pictures of the authors, and user testimonials. It also highlights Suns vague assurances that this (if it lives up to the hype) amazing utility may or may not end up in the public domain."

212 comments

  1. assurance of what ??? by mirko · · Score: 3, Funny

    It also highlights Suns vague assurances that this amazing utility may or may not end up in the public domain."

    so, we are assured of a and !a ???

    --
    Trolling using another account since 2005.
    1. Re:assurance of what ??? by Anonymous Coward · · Score: 4, Funny
      so, we are assured of a and !a ???

      No, we're assured of a or !a.

      Although knowing Sun, they're probably trying for a and !a and wondering why it won't work...

    2. Re:assurance of what ??? by Cardinal+Biggles · · Score: 0

      so, we are assured of a and !a ???

      a or not a, actually. No need to be vague about that assurance. I will even assure you that for any a. ;-)

    3. Re:assurance of what ??? by 4lex · · Score: 2, Funny

      It also highlights Suns vague assurances that this amazing utility may or may not end up in the public domain." so, we are assured of a and !a ???

      No, we are _vaguely_ assured of a *or* !a. I thought Sun made it clear enough! ;)

      --
      My journal. Mainly about freedom.
    4. Re:assurance of what ??? by meringuoid · · Score: 1
      a or not a, actually. No need to be vague about that assurance. I will even assure you that for any a.

      I hereby define statement a as:

      Statement a is false.

      a, or not a, that is the question...

      --
      Real Daleks don't climb stairs - they level the building.
    5. Re:assurance of what ??? by NuclearDog · · Score: 0

      While we're on the topic of true or false statements:

      The statement below this is true.
      The statement above this is false.


      ND

      --
      This statement is forty-five characters long.
  2. And what is DTrace? by Anonymous Coward · · Score: 4, Insightful

    It looks like it's an allround system monitoring and administration tool. But how is it different/better than a well-sorted collection of individual tools, each doing one job as good as possible?

    1. Re:And what is DTrace? by davecb · · Score: 5, Informative

      It's a mechanism for adding trace calls to pretty-nearly-arbitrary locations.

      --
      davecb@spamcop.net
    2. Re:And what is DTrace? by Anonymous Coward · · Score: 0

      No one can tell you what DTrace is, you have to see it for yourself.

    3. Re:And what is DTrace? by int19 · · Score: 5, Informative

      It is much, much more than that. The problem I find with a "well-sorted collection of individual tools" in this case is the complexity/difficulty/impossibility to do what dtrace can do. Check out this usenet posting for some examples.

      I've been using it for a little while, it's a very interesting tool.

    4. Re:And what is DTrace? by afidel · · Score: 4, Interesting

      VERY cool indeed. Seems better than any tool for any OS I've seen. The closest would be process explorer from sysinternals but the level of detail is nowhere near what dtrace provides. For instance one thing that I've never been able to figure out is how to backtrace a file lock to the owner process.

      --
      There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
    5. Re:And what is DTrace? by chegosaurus · · Score: 4, Informative

      That's kind of like saying perl is an all round text processing tool, then asking why using perl is better than using cut, sort, and tr.

      You can do a lot with cut sort and tr. Often they're all you need, but perl lets you solve problems those three tools can't even address.

      I'd recommend grabbing a Solaris 10 beta and having a play with this thing. It's pretty amazing.

    6. Re:And what is DTrace? by Anonymous Coward · · Score: 0

      > For instance one thing that I've never been able to figure out is how to backtrace a file lock to the owner process.

      Process Explorer will find who has a handle to the file. That's probably also the one who locked it.

    7. Re:And what is DTrace? by afidel · · Score: 1

      Yeah except you can't search by handle so on a busy file or app server it's basically worthless.

      --
      There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
    8. Re:And what is DTrace? by christophersaul · · Score: 2, Insightful

      Reading the article would help you find out.

    9. Re:And what is DTrace? by 4of12 · · Score: 1

      So how about this wild ass attempt for an explanation:

      Is dtrace simply a way of inserting wrapped, more-instrumented system calls to replace the standard ones, on the fly, perhaps using the same concepts as dynamic loaded libraries?

      [I'm accustomed to memory debuggers that work by substituting their own heavily instrumented wrapper versions of malloc() for the vanilla ones.]

      --
      "Provided by the management for your protection."
    10. Re:And what is DTrace? by upsidedown_duck · · Score: 1

      For instance one thing that I've never been able to figure out is how to backtrace a file lock to the owner process.

      man fuser

      Also, for the masochistic out there, it is possible to look at the pfiles output for every process and match inodes.

      --
      -- "Makes Little Debbie look like a pile of puke!" - Moe Szyslak
    11. Re:And what is DTrace? by Anonymous Coward · · Score: 0

      Try this:
      http://www.sysinternals.com/ntw2k/freeware/ handle. shtml

    12. Re:And what is DTrace? by afidel · · Score: 1

      Gracias senior.

      --
      There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
  3. It's about time (;-)) by davecb · · Score: 4, Interesting

    Dtrace is a lovely mechanism, and once it's out in production Solaris (and maybe Linux) I'm going to write a new TPS/response-time monitor using it.

    --
    davecb@spamcop.net
    1. Re:It's about time (;-)) by Anonymous Coward · · Score: 0

      Did you get the memo that we're sending out all our TPS reports with the new cover sheet?

    2. Re:It's about time (;-)) by int19 · · Score: 3, Funny

      Yes, I got them memo, and I've already taken care of it ... so ... it's not really a problem anymore.

      ;-)

    3. Re:It's about time (;-)) by little_fluffy_clouds · · Score: 3, Funny

      Yeah. It's just we are putting new coversheets on the TPS reports before they go out, so if you could go ahead and try to remember to do that from now on, that would be great.

      --
      What were the skies like when you were young?
    4. Re:It's about time (;-)) by sql*kitten · · Score: 4, Funny

      going to write a new TPS/response-time monitor using it.

      Did you get the memo about the cover sheet?

    5. Re:It's about time (;-)) by Anonymous Coward · · Score: 0

      Hey int19? What's this I hear about a problem with your TPS reports? Didn't you get the Memo? I'll tell you what, let me send you a copy of the memo again. ok?

    6. Re:It's about time (;-)) by [tsa] · · Score: 0, Offtopic

      Who cares about Linux, anyway? Solaris rocks,
      especially with dtrace/zfs/zones/..

    7. Re:It's about time (;-)) by Surt · · Score: 1

      When i was young the sky was mostly brown and orange. Now it is mostly blue. A vast improvement in the air quality has really happened over the years, and it looks like things are going to continue to improve!

      --
      "Who is the Journal of Quantum Physics going to believe?" --Stephen Hawking
    8. Re:It's about time (;-)) by Anonymous Coward · · Score: 0

      Sweet! Now we can all finish our TPS reports

    9. Re:It's about time (;-)) by Almost-Retired · · Score: 1

      When i was young the sky was mostly brown and orange. Now it is mostly blue. A vast improvement in the air quality has really happened over the years, and it looks like things are going to continue to improve!

      Hummph. Youngster Whippersnapper. I was here when they didn't need any improvement.

      When I was young we had blue skys unless it was raining, and clear, dark skys at night, so dark and clear you could see the milky way from horizon to horizon. The "north star", Polaris, could be picked out of the crowd of bright, reach up and cut your hand on them, stars in about 1 second flat on the average night. Vega, which will be our north star in another 23,000 years, was just as easy to find.

      This was in the early 1940's, before the beloved automobile shot its wad into our air, seemingly forever, although its getting better the last decade or so again.

      So you want to see the milky way in all its glory? Easy, have Burt take you up in space ship 1, but you aren't gonna see it from ground level till we really, really clean up the air, and put some real dark skys laws into effect for nighttime lighting of parking lots etc.

      That will put you back to reading by kerosene lamps, and listening to the radio on crystal sets, but with modern ccfl lighting, and solid state entertainment that doesn't need a 500 watt 5.1 surround sound and cannot destroy your ears, we wouldn't be far from cleaning up the air. Oh, and I forgot, you'd be riding old Nelly the mare to school, but since it was 2 miles to school, and the last decent barn to leave old Nelly in for the day in bad weather was only a mile away, you'd walk that last mile to school in bad weather.

      Been there, done that. And I damned sure miss those clear, cold as hell skys we had back then.

      Cheers, Gene

  4. terrrible article by XMichael · · Score: 4, Insightful

    This article seems like it was written by Sun. I bet it's one of them junket articles, Sun must be a major sponsor of theregister.co.uk or something...

    The article says what this D-Trace does, but yet, doesn't... Not to mention it's neither objective, review like, and even fails to mention alternatives or relative tools

    Definately one of the poorest software article's I've seen in a while

    Slashdot is making me bitter (-;
    Complete CCTV Security Cameras

    1. Re:terrrible article by mpcooke3 · · Score: 1, Funny

      At least with thereg first you get the poor reporting first hand :)

    2. Re:terrrible article by chamcham · · Score: 0
      The article says what this D-Trace does, but yet, doesn't...

      Journalists are known for their vast technical knowledge.

    3. Re:terrrible article by JPelorat · · Score: 0, Troll

      I was waiting for the paragraph where it described DTrace's ability to cook breakfast, make coffee, and suck cock all at the same time.

      --
      Hokey statistics and ancient misconceptions are no match for a good thought in your head, kid!
    4. Re:terrrible article by Handyman · · Score: 4, Insightful

      Sun must be a major sponsor of theregister.co.uk or something...

      I must agree that the article is pretty positive about the whole thing. But a positive review is not necessarily a bought review. In fact, the article does contain an angle that is not likely to be sponsored by Sun: open sourcing it, and open sourcing solaris. They explicitly make fun of Sun's president, who is evading their questions about open sourcing DTrace. From the article:

      So there you have it. DTrace may or may not end up in the public domain. Glad that's settled.

    5. Re:terrrible article by Anonymous Coward · · Score: 3, Informative

      The article says what this D-Trace does, but yet, doesn't..
      The article gives sample uses, but it isn't a howto. I don't usually go to the register for technical howto's, do you? The article does link to more info on how to use the utility.

      Not to mention it's neither objective,
      Nor is it wholy pro-sun, it jabs the rep for not directly saying whether or not it will be open sourced and repeatedly makes comments about the hype ("It slices, it dices, it spins, it whirls").

      and even fails to mention alternatives or relative tools
      It mentions that various system utilities could be used to do somewhat the same thing, on a much different level, one of the testimonials mentions that a debugger could have been used too, but would have been slower and another user mentions using prex to get similar information.

    6. Re:terrrible article by the+chao+goes+mu · · Score: 2, Funny

      At least it doesn't say Dtrace was a scalable,, three-tier, object-oriented, java-driven solution engine...

      --
      Boys from the City. Not yet caught by the Whirlwind of Progress. Feed soda pop to the thirsty pigs.
    7. Re:terrrible article by the+chao+goes+mu · · Score: 0, Offtopic
      I seem to be having a problem with tenses in my last post:

      s/was/is/

      --
      Boys from the City. Not yet caught by the Whirlwind of Progress. Feed soda pop to the thirsty pigs.
    8. Re:terrrible article by Hubert_Shrump · · Score: 2, Funny

      you're thinking of longhorn.

      --
      Keep your packets off my GNU/Girlfriend!
    9. Re:terrrible article by [tsa] · · Score: 1

      Uhm...maybe because there are no alternatives so
      far?

    10. Re:terrrible article by supersnail · · Score: 1

      I agree with this. The article was very uncritical and did not place the software in any context.

      IBM mainframe users have had this type of stuff since the early '80s. (The trace facilities SMF etc. were much earlier but the cool tools for analysing the output weren't really there util then).

      Also there are similar tracing facilities available on AIX, although the user interface is classic IBM (obscure and picky) syntax.

      This could be viewed as case of Sun playing catchup with its competitors, and, hyping up any new product to obscure the fact that it hasn't any new hardware to sell.

      --
      Old COBOL programmers never die. They just code in C.
    11. Re:terrrible article by Kalak · · Score: 1

      No one ever said The Register can't be bought

      --
      I am, and always will be, an idiot. Karma: Coma (mostly effected by .hack)
    12. Re:terrrible article by upsidedown_duck · · Score: 1

      Definately one of the poorest software article's I've seen in a while

      There was another article a while ago at osnews.com about DTrace. If I recall, one of Sun's engineers was responding in the forum. I haven't used DTrace, yet, but there is a lot of non-Sun positive responses out there to help corroberate the Register article.

      --
      -- "Makes Little Debbie look like a pile of puke!" - Moe Szyslak
  5. indeed! by rylin · · Score: 3, Funny

    Bingo!
    In fact, you can bet your life that it will or won't show up in public domain.
    Now this is what I call a nice and reputable company! At least they keep what they promise!

    1. Re:indeed! by Anonymous Coward · · Score: 0

      you can bet your life that it will or won't show up in public domain

      What if only parts of it are open sourced, so it manages to both show up AND not show up in public domain? So much for your life!

  6. Re:Lots of users == good. by REBloomfield · · Score: 1

    So you're using Solaris 10 then, yeah?

  7. Very Useful by trifakir · · Score: 2, Interesting

    Sounds great and I am quite impatient to see how it works (take a mental note to increase the priority of the Solaris upgrade on one of my toy SPARCs). I can see a growing need for a non-obtrusive profiling tool. As a matter of fact I started writing something which may be similar to dtrace for Linux and it is for a long time in my TODO to port it to Solaris and maybe IRIX.

    1. Re:Very Useful by dpb · · Score: 1

      Urm, its already been done by IBM a long time ago. They ported DProbes from OS/2 to Linux way back in 2000. SuSE has distributed it for quite a while & some parts are now in stock 2.6 kernels. See the IBM DProbes project

    2. Re:Very Useful by Anonymous Coward · · Score: 0
      Here are some dtrace features which DProbes lacks:
      • Thread local variables
      • Associative arrays
      • Data aggregation
      • Predicate support
      • Safety (i.e. a bad dprobe can crash the system)
    3. Re:Very Useful by kscguru · · Score: 4, Informative
      Okay, since you've felt the need to hype DProbes at least a couple of times, I'll pull from the DTrace USENIX paper (link was somewhere up above).

      The major difference: a huge amount of polish. DProbes can crash your system if used incorrectly; DTrace is guaranteed safe. DProbes reports just about everything; DTrace supports speculative tracing and predicates so you can narrow down the fire-hose of profiling data into the useful information. DProbes doesn't scale well across multiple CPUs; DTrace does. (This last one is just like Linux versus Solaris: one scales, the other doesn't. Sorry!)

      So, yes. If IBM had spent the past four years tuning DProbes and turning it into a useful tool, they might have something. As it is, Sun spent quite a bit of time and money, and ACTUALLY CREATED a very powerful tool.

      [rant]What is increasingly bugging me about open source advocates on Slashdot is how they'll look at any (ANY!) closed-source tool and say "oh, open-source tool XYZ does this / did this a few years ago / will make you walk on water". These advocates will cheerfully ignore that the closed-source tool has months or years of additional polish that have brought it from an internal company toy project into a commercially viable tool - that the open source tool of now is no more advanced than the closed source tool of years ago, except that most open source tools are stagnant and will never leave development and never obtain the degree of polish wide-scale use demands.

      A message to open-source advocates: if you don't spend the time turning your hobbyist code into solid, professional code, the closed-source companies that DO spend the time will leave you in the dust - freedom of code (both free as in beer and free as in GNU) notwithstanding.[/rant]

      --

      A witty [sig] proves nothing. --Voltaire

    4. Re:Very Useful by msconception · · Score: 1

      Actually, the dtrace concept and functionality is technology that was around about 20+ years ago in a Sperry OS (now Unisys) called VS/9. It was known as IDA (Interactive Debugging AID) and was much more user-friendly and powerful than dtrace. However, it was not for Solaris. IDA was designed integrated into VS/9 as a command language from day one - not some afterthought after a multitude of tools existed. Therefore, it's monitoring commands could be scripted into regular job control.
      dtrace has potential if it is provided a more user-friendly interface. If not, it's essentially one step up from writing a c program to OS probe system calls. Not very inventive.
      It's interesting how Sun feels the need to reinvent rather than embrace and augment existing tools - just adding to a confusing mix of paths of "which tool is best for which task". Why not provide a be-all tool with user friendliness to encompass user and sysadmin tasks that eleminates need for any other tools? Why the sprinkling of tools that don't quite do everything?
      By the way, /IDA could crash a program when used by a user and/or the OS when used by the superuser - features that were sometimes desired but had to be enabled via an option.
      As for creating a powerful tool for public use: I suspect dtrace was built for internal use and determined to be good enough for public. It has the user-unfriendly design feel as such.

  8. other resource by SubtleNuance · · Score: 5, Informative

    An excellent article at sys admin mag.

  9. Re:Lots of users == good. by stripyd · · Score: 1

    I think pkgadd may be the command you're looking for squire :-)

  10. just to save you all some trouble by chegosaurus · · Score: 5, Insightful

    No, linux does not already do this.

    Solaris users could not care less whether Sun ports it to linux or not.

    Sun are *not* evil because they don't immediately give away the source to a product which has taken them years and $$$s to develop and which gives them an edge in a competetive market.

    Some people just prefer Solaris. If you prefer linux, that's fine.

    1. Re:just to save you all some trouble by Anonymous Coward · · Score: 0

      Some people use both. Why the snooty comment?

      Because there are already people whinging that doesn't exist for Linux. And we don't need that.

    2. Re:just to save you all some trouble by chegosaurus · · Score: 4, Insightful

      Wasn't meant to sound superior! I use both (and OpenBSD) myself. I'm expecting this thread to be full of ill-informed "linux does this Slowaris is teh sux" posts and whining about "why won't they GPL it it"?

      So I guess I'm saying, if you're 100% linux, that's cool. I certainly don't need to hear about j00r leet gentoo boxen in a Solaris thread. You know the kind of person I mean!

    3. Re:just to save you all some trouble by Florian+Weimer · · Score: 0, Troll

      No, linux does not already do this.

      Show me a tool that can link high disk I/O load to the processes that cause it, on a vanilla Linux kernel.

    4. Re:just to save you all some trouble by Anonymous Coward · · Score: 0

      If they port it, fine.

      If they don't, someone will write an opensource program that does the same thing.

    5. Re:just to save you all some trouble by Anonymous Coward · · Score: 0

      > No, linux does not already do this.

      Show me a tool that can link high disk I/O load to the processes that cause it, on a vanilla Linux kernel.


      Um, yes, that's precisely what he said? ;)

    6. Re:just to save you all some trouble by Anonymous Coward · · Score: 0

      You know the kind of person I mean!

      Ja.

      It's kind of like those persons who decide to sweep people into broad, stereotyped groups and deride them because the derider is superior in their ability to refrain from 'teh l33t sp34kz' while running 'teh gentoo b0xen'.

    7. Re:just to save you all some trouble by dpb · · Score: 4, Informative

      > No, linux does not already do this.

      Yes, it does (well, 85% of it at least), in the form of DProbes. Dynamic Probes was ported by IBM from OS/2 to Linux way back in 2000, is distributed by SuSE and some parts are in 2.6 kernel out-of-box.

    8. Re:just to save you all some trouble by DAldredge · · Score: 1

      The PP said that such a tool does not exist...

  11. Huh? Quantum leap? by Anonymous Coward · · Score: 1, Funny

    "DTrace is a quantum leap over previous tools"

    Doesn't that mean it is a very small, randomly occuring change? Silly marketing speak.

    1. Re:Huh? Quantum leap? by sarabob · · Score: 3, Funny

      No, he means that it's a relatively large change with no intermediate positions.

      "very small" is only your perception of the size of change. Moving between a base state and an excited state is a pretty big thing for an atom.

    2. Re:Huh? Quantum leap? by Darkfred · · Score: 2, Informative

      I think you do not have a good understanding of the meaning of the word quantum.
      Quantum means 'a level, or fixed amount'. In Quantum mechanics this refers to the smallest observable physical effects. At which point changes in energy are in fixed 'Levels' (ohh like Qunatums) rather than smooth functions.

      So in this article he is saying that this has reached the next level. And that you are a dimwit.

      Thank you very much,
      QED

      --
      ----- 70% of all statistics are completely made up.
    3. Re:Huh? Quantum leap? by cluke · · Score: 1, Interesting

      Dictionary.com diagrees with you sir!
      "A dramatic advance, especially in knowledge or method...This term originated as quantum jump in the mid-1900s in physics, where it denotes a sudden change from one energy state to another within an atom. Within a decade it was transferred to other advances, not necessarily sudden but very important ones."

      (As a side note, I did a google on quantum leap definition and got a lot of results using the example sentence ""this may not insure success but it will represent a quantum leap from last summer"
      The exact same sentence! Which if my knowledge of English serves, incorrectly uses 'insure' where 'ensure' was intended. Am I wrong? Either way, just goes to show the Internet is just a mass of plagiarism)

    4. Re:Huh? Quantum leap? by AliasTheRoot · · Score: 2, Funny

      If it doesn't make me say "Oh boy..." in a resigned kind of way it is no Quantum leap.

    5. Re:Huh? Quantum leap? by malkavian · · Score: 2, Funny

      As opposed to humans, who usually exhibit more of their base states when excited.

    6. Re:Huh? Quantum leap? by the+chao+goes+mu · · Score: 1

      Or a change which sends Brian Benben through a series of unrelated mediocre melodramas spread throughout several decades...

      --
      Boys from the City. Not yet caught by the Whirlwind of Progress. Feed soda pop to the thirsty pigs.
    7. Re:Huh? Quantum leap? by grey1 · · Score: 1

      re insure/ensure - yup, I'd say it should be "ensure"; maybe an American can comment on the American English view of this.

      G

      --
      "we demand rigidly defined areas of doubt and uncertainty!"
    8. Re:Huh? Quantum leap? by Anonymous Coward · · Score: 0

      I think you mean Scott Bakula.

    9. Re:Huh? Quantum leap? by Anonymous Coward · · Score: 0

      Thank you very much,
      QED


      Quantum Electrodynamics?

    10. Re:Huh? Quantum leap? by Anonymous Coward · · Score: 0

      Insure and ensure mean the same thing, it is only current usage which suggests otherwise.

  12. Isn't this a profiler for non-profiled binaries? by BusterB · · Score: 0

    So, one guy uses this tool to analyse function calls in a program. Other times, it is used to analyse whether tools like Oracle or BEA are making system calls.

    This is exactly what a profiler is for. Doesn't Solaris come with profiled libraries you can link your code to to do the same thing with something like gprof? Aside from the fact that this tool works with non-profiled binaries, I fail to see the revolutionary aspect.

  13. Dtrace slashdot by hhg · · Score: 3, Funny

    I bet someone has to DTrace his slashdotted server real soon now..

  14. Re:Lots of users == good. by mollymoo · · Score: 4, Funny
    If I can't use it, then I'll never learn about how good it is.

    So if some software isn't in the Gentoo portage tree you won't run it on your Solaris boxes at work?

    This is proprietary software for a version of Solaris which isn't even out yet.

    Remind me not to employ you as a Solaris sysadmin. Hell, remind me not to employ you to make the tea - we might have different cups at work and you wouldn't know what to do with them.

    --
    Chernobyl 'not a wildlife haven' - BBC News
  15. a || !a = 1 by pjt33 · · Score: 2, Funny

    No, we're vaguely assured that [i]true[/i].

  16. i/o preformance tuning by xyote · · Score: 4, Informative
    I've used stuff that gave you a realtime display of i/o perfmance stats and you could see when the hotpoints were happening. The problem was how to fix it. It's not like the old days when you had maybe 3 disks and you just moved a filesystem or database partition to another disk. Now you have huge amounts of raid storage with logical volume managers and your logical volumes (partitions) are all over the place. It's not so easy. You could see a hot spot and determine that it's caused by two logical volumes sharing a physical drive. Except when you move one to a different drive, you could create different hot spots and they could be much worse. Kind of like wack-a-mole except you're playing it while you're on call. Not fun.

    It's a hugely complicated problem and it doesn't have a simple solution except in the minds of markedroids.

    1. Re:i/o preformance tuning by TheTomcat · · Score: 1

      Pardon my ignorance, but if you are able to monitor i/o (and hotspots), on a full system, wouldn't it be possible to estimate/simulate what would happen if you moved a logical volume to another physical disk?

      1) monitor for a period of time (1 day, one week, something typical that sees all normal usage patterns);
      2) analyze collected data;
      3) simulate moved (physical) data;
      4) goto 2

      Perhaps I'm missing something -- I realize this would be complex, but does not sound impossible.

      S

    2. Re:i/o preformance tuning by kbahey · · Score: 1

      I hear ya!

      And it is not only RAID and Logical Volumes locally on the box.

      Things get more complex when you have things like EMC Symmetrix storage with more degrees of indirection inside it.

      I/O has become more complex over the years, and some things that were specifically done to avoid I/O contention in in UNIX/Linux and in databases (e.g. Oracle) are no longer relevant.

      Things have come a long way.

  17. looking glass by stripyd · · Score: 3, Interesting

    I'm not sure how much of an indicator the open sourcing of looking glass is for what Sun decide to do with dtrace. Looking Glass never seemed to be anything more than a nice piece of eye candy to showcase the java desktop (the real product they were flogging). dtrace on the other hand looks like being one of their biggest pieces of product differentiation for Solaris in years. We all know how finance departments like those transactions-per-second->cost-per-transaction figures in making procurement decisions...

    I suppose we'll have to wait and see...

    1. Re:looking glass by dmiller · · Score: 1

      Why don't you stop expecting Sun to release it and write a replacement for your_favourite_free_os? Free software doesn't exist because of scraps from the corporate table, but because people get off their arses and write good software.

    2. Re:looking glass by stripyd · · Score: 1
      Why don't you stop expecting Sun to release it...

      That post...I do not think it means what you think it means. If you read it again you may find that I am simply stating that despite suggestions that the open sourcing of looking glass might be a precedent to open sourcing dtrace, the economics of the two actions are different due to the different commrecial potential of the products.

      I expect nothing more from Sun than to be fair to their employees (of which I used to be one a decade ago), adhere to standards, not engage in overly dodgy business practices, and continue to provide excellent operating systems which are worth paying for. An aspect of the former expectation is preserving their competetive advantage over other companies.

      I understand and sympathise with the point you're making, but you're replying to the wrong post.

      FWIW, just about all the userland code I've got off my arse and written in the last few years compiles on Linux and Solaris despite Sun's pam implementation :-)

  18. Re:Lots of users == good. by caluml · · Score: 1

    Grr. No-one understands what I meant.

  19. Guaranteed by Anonymous Coward · · Score: 0

    With statements like

    "assurances that this (if it lives up to the hype) amazing utility may or may not end up in the public domain."

    how can we go wrong. :-)

  20. Microsoft perfmon by esarjeant · · Score: 3, Interesting

    Not to be the devils advocate, but Microsoft has offered excellent detailed profile management via perfmon for quite some time now. You can inspect disk IO per-processes, memory per-processes, page faults, etc. etc.

    It's a bit of a misnomer to characterize dtrace as the first application for doing this, although I will say it's among the first that I have seen that is scriptable (a huge advantage when troubleshooting problems remotely).

    --

    Eric Sarjeant
    eric[@]sarjeant.com

    1. Re:Microsoft perfmon by Anonymous Coward · · Score: 2, Informative

      DTrace does much, much more than what perfmon does. They aren't even close to the same level in terms of their capabilities. And I say this, while being Windows developer myself..

    2. Re:Microsoft perfmon by AliasTheRoot · · Score: 2, Interesting

      It's been a while since I adminned any Solaris boxes in anger, but there was stuff on them that was equivalent or better than perfmon, and this was back with 2.7 / 2.8 machines.

    3. Re:Microsoft perfmon by Anonymous Coward · · Score: 1, Informative

      DTrace is more of a language than an actual app...

      Check out http://www.sun.com/bigadmin/content/dtrace - has more info about it.

    4. Re:Microsoft perfmon by Anonymous Coward · · Score: 0

      Dude .. this isnt per process anything .. Dtrace is much more fine grained .. can tell you which functions anbd system calls that are causing problems .. in a few lines of dtrace script you can print out a pretty graph of your top 10 i/o consumers etc.

      Knowing a process is a resource hog isn't much use when you are trying to fix a problem on a production server on an app you need. (You cant just kill the process).

    5. Re:Microsoft perfmon by Zode · · Score: 5, Insightful

      are you kidding? DTrace lets you do this not just per process, but per instruction, across CPUs. Comparing DTrace to perfmon is like comparing a Ferrari to a tricycle.

  21. Re:Lots of users == good. by Anonymous Coward · · Score: 0
    With an attitude such as yours (if I can't run it on Gentoo then I won't bother learning about it), chances are, you won't make it to decision maker status.


    Tunnel vision on a sysadmin, (I don't care if you're taking care of Windows, Linux, Solaris, BSD, or Mac for fucksake) is a bad thing [tm].

  22. Sun reminds me of H2G2 philosophers by arvindn · · Score: 4, Funny

    Sun's vacillation about open source reminds me of the philosophers in the Hitchhiker's guide.


    ``But who the devil are you?'' exclaimed an outraged Fook.

    ``We,'' said Majikthise, ``are Philosophers.''

    ``Though we may not be,'' said Vroomfondel waving a warning finger at the programmers.

    ``Yes we are,'' insisted Majikthise. ``We are quite definitely here as representatives of the Amalgamated Union of Philosophers, Sages, Luminaries and Other Thinking Persons, and we want this machine off, and we want it off now!''

    ``What's the problem?'' said Lunkwill.

    ``I'll tell you what the problem is mate,'' said Majikthise, ``demarcation, that's the problem!''

    ``We demand,'' yelled Vroomfondel, ``that demarcation may or may not be the problem!''

    ``You just let the machines get on with the adding up,'' warned Majikthise, ``and we'll take care of the eternal verities thank you very much. You want to check your legal position you do mate. Under law the Quest for Ultimate Truth is quite clearly the inalienable prerogative of your working thinkers. Any bloody machine goes and actually finds it and we're straight out of a job aren't we? I mean what's the use of our sitting up half the night arguing that there may or may not be a God if this machine only goes and gives us his bleeding phone number the next morning?''

    ``That's right!'' shouted Vroomfondel, ``we demand rigidly defined areas of doubt and uncertainty!''

    1. Re:Sun reminds me of H2G2 philosophers by Anonymous Coward · · Score: 0

      ``I mean what's the use of our sitting up half the night arguing that there may or may not be a God if this machine only goes and gives us his bleeding phone number the next morning?''

      Hey, wait a minute...

      (dials 42)

      Hi, uh, I'd like to speak to the manager please.

      No, I don't want to return any faulty commandments. YES, I know you don't take prayers back either, I just want a chat with the boss, okay?

      What? No, no I'm not a regular churchgoer...

      Oh, for Christ's sake, will you just put me through!

      Eh? What do you mean, blasphem---(zzzzap!!!)

      NO CARRIER

    2. Re:Sun reminds me of H2G2 philosophers by Anonymous Coward · · Score: 1, Funny

      Slashdot rule #321: HHGTTG, Simpsons, etc. quotes don't actually have to make sense or have context in the current discussion.

    3. Re:Sun reminds me of H2G2 philosophers by Cryptnotic · · Score: 1
      Slashdot rule #321: HHGTTG, Simpsons, etc. quotes don't actually have to make sense or have context in the current discussion.


      Not even to be modded up to +5 funny, aparrantly.
      --
      My other first post is car post.
  23. Interesting... by starseeker · · Score: 5, Interesting

    They seem to have a lot of praise from users, not just market speak. And technical users yet.

    I'd be very, VERY surprised if Sun allows Dtrace into the open source world, at least not for a while. If Dtrace really is the supertool it seems to be, and is actually and massively UNIQUE, it represents a reason peole will move to Solaris and buy Sun's hardware to do it. Maybe the closest thing to a Unix killer app that has existed for a while.

    Now eventually (as in five years down the road) it will probably pay for Sun to open it up. If I were them, I'd milk it for all its worth on the "get people to move to Sun boxes" mantra, while the rest of the world trys (and probably fails) to duplicate the tool. Then, when Sun has gotten all the converts they are likely to, start making the tool even BETTER by opening it up and letting the world go to town on it. (GPL or something similar so Sun can incorporate back in the goodies.)

    Of course, that's just an off the cuff theory by someone who doesn't know.

    --
    "I object to doing things that computers can do." -- Olin Shivers, lispers.org
    1. Re:Interesting... by Florian+Weimer · · Score: 1

      I'd be very, VERY surprised if Sun allows Dtrace into the open source world, at least not for a while.

      Oddly enough, there doesn't seem to be a patent for DTrace.

    2. Re:Interesting... by gr8_phk · · Score: 3, Insightful
      "Maybe the closest thing to a Unix killer app that has existed for a while."

      Ya, sysadmins are always trying to optimize code in the core applications they bought from vendors. NOT.

      It's not useless, but the examples cited were from application developers optimizing their code.

    3. Re:Interesting... by mindstrm · · Score: 1

      Really.

      Because I can think of a number of applications in my organization where it would be GREAT for me to be able to definitively pinpoint where the bottlenecks are, instead of spending weeks going back and forth between vendors.

    4. Re:Interesting... by Keith+Maniac · · Score: 1

      It's not just about optimizing code, it's about understanding what the code is doing.

      Whether you wrte the code or not (or even have source), sometimes the results are complicated. Why is this app so slow?

      Is it disk I/O wait? You can check that. Is the dumbass programmer opening thousands of little files and throwing them away? Sure, you can check that. Are you paging, but only when some other app starts grinding away?

      It's *not* a tool to optimize code. It's a tool to understand the behavior and problems with running programs. The fact that you can use that info to optimize your application code is just a bonus.

    5. Re:Interesting... by _damnit_ · · Score: 1

      I think you are focusing too narrowly on this. The impact COULD be on many fronts.
      First, as a sysadmin you could determine where your problems exist and provide proof to your internal cevelopment teams or to the vendor. I would find this priceless as most of us have been on the vendor seesaw before. Oracle blames SAP who Oracle and then both intimate it could really be the OS or Veritas...
      Second, app developers can use this tool to dramatically improve the performance of their apps. You don't think db vendors will use this to improve their apps to gain a benchmark advantage over competitors? Some Solaris gains could even improve the performance of apps on other OSes.
      Third, Sun support will use this extensively to determine where customer problems lie and get real root causes much more often without live coredumps and the like. Most customers will be very pleased with being able to get a detailed view of their boxen while the problem is occuring and a faster resolution.

      There are other ways this could help Solaris users when you add in the Solaris Zones, but I'll leave that to the imagination. All in all, there are many reasons Sun should be very happy with this and sing its praises to the world. The real issue is that most people do not move to new OSes quickly. There are LOTS of people still running 2.6. Heck, a lot of big customers have probably just finished moving to Solaris 8 (2.8, SunOS 5.8). There have to be a lot of really good reasons to move to Solaris 10 for companies to shift directions and start deploying. In this case, I believe there are. The Solaris Express program should help a lot as well.

      DISCLAIMER: I work for Sun. I do not drink the company Kool-Aide though.

      --


      _damnit_

      It's my job to freeze you. -- Logan's Run
    6. Re:Interesting... by TheLastUser · · Score: 1

      Solaris already has tons of reasons to choose it over linux. Its only failing is that is has not completely adopted and integrated the gnu tools. Once they dump all of that old system V shit and replace with gnu, they will have a killer OS. Especially if they conform the layout as much as possible to the Linux standard.

  24. Re:Isn't this a profiler for non-profiled binaries by Anonymous Coward · · Score: 0

    Sun's "collect" program that ships with their compiler suite already does this. No special compile or link step is necessary.

  25. Understanding and using Dtrace by ChrisRijk · · Score: 5, Informative

    Dynamic Instrumentation of Production Systems - this paper was presented at Usenix, and describes how Dtrace is actually implemented.

    Dtrace user guide.

    A collection of Dtrace scripts

    1. Re:Understanding and using Dtrace by Frisky070802 · · Score: 1

      Yes, and from seeing the USENIX presentation I can understand why the article referred to Cantrill as "energetic". After his presentation, the next speaker even joked about having to live up to the high standard that had just been set. If USENIX gave a "best presentation" award I suspect that talk would have won hands down.

      --
      Mencken had it right. So glad that's old news.
  26. Re:Lots of users == good. by Anonymous Coward · · Score: 2, Informative

    Grr. No-one understands what I meant.

    Which is what - you don't have an opportunity to play with it at home? But you do - you can get a Solaris 10 preview through their Sun Express program (linked in the article) and install it on another box, in VMWare, etc.

  27. Re:Lots of users == good. by sarabob · · Score: 0

    So what did you mean then?

    That emerge should give you a "why you should use dtrace" document?

  28. Kprobes and Dprobes by Anonymous Coward · · Score: 4, Interesting


    How does Dtrace for Solaris differ from Kprobes and Dprobes for Linux ??

    1. Re:Kprobes and Dprobes by chegosaurus · · Score: 1

      Dprobes is discussed (briefly) in the announcement paper. (PDF - sorry.)

    2. Re:Kprobes and Dprobes by Dug · · Score: 5, Informative
      The Usenix paper has some detail on the differences.

      Dtrace Usenix Paper

      More examples in answerbook from the BigAdmin Dtrace page

    3. Re:Kprobes and Dprobes by Anonymous Coward · · Score: 5, Interesting

      The mechanism (dynamic instrumentation) is similar. What makes DTrace unique over and above k/dprobes is the incredible power of the 'D' script language. It features things such as speculations (look at the docs on sun.com) that are absolutely vital in many situations for diagnosing problems - too much data is often as bad as too little data, and DTrace lets you easily get exactly the data you want, and only the data you want.

      You can find a small section comparing kprobes with DTrace in the DTrace USENIX paper; it also mentions some other differences (kprobes can easily crash your production system; that won't happen with DTrace).

  29. Tru64 had this years ago by decsnake · · Score: 1

    Maybe Compaq patented it? HP could sue Sun. Wouldn't that be fun?

  30. Re:Hmmm by Anonymous Coward · · Score: 3, Funny

    Give system catchy name, vaguely reminiscent of Eminem's rap group

    I'm sure they did this to appeal to the disproportionately-large numbers of African-Americans in the field of computer science.

  31. New Tool? Maybe for Solaris... by Procrastin8er · · Score: 2, Informative

    Not sure what all the hype is about, DTrace sounds alot like HP's Measureware and PerfView. Bundling it with the OS for no addition cost, that sounds good to me.

    --
    Slashdot - Where the slash is most definitely to the left.
    1. Re:New Tool? Maybe for Solaris... by Anonymous Coward · · Score: 0

      No, PerfView and Measureware are just high level aggregators and displayers of standard unix performance tools.

    2. Re:New Tool? Maybe for Solaris... by Anonymous Coward · · Score: 0

      I'm glad someone else here is familiar with HPUX, or at least MeasureWare.

      For those not in the know, MeasureWare (a.k.a. OpenView Permormance Manager), has been around for quite a while and is available for a number of platforms, including some x86 Linux distros. There are numerous pieces to it, starting with the measureware daemon, which actually does the measuring. Then there are add-ons, that can use that data, like Glance and GlancePlus, PerfView, and OpenView reporting server.

      Under HPUX, the kernel makes a huge number of statistics available to the measureware daemon, which collects them every 3 seconds. These are not snap-shot like samples, like sar. The data collected actually contains all of the kernel statistics generated over that time period and because it is provided by the kernel itself, it takes insignificant amouns of CPU overhead.

      Under other operating systems, like Solaris and Linux, the measureware daemon collects the data much more like ps, top, vm_stat or other more traditional monitoring tools and therefor takes slightly more overhead.

      Once the data has been sampled, the measureware daemon, records it to the HD acording to the config file.

      The other tools can look at both the "live" and recorded data, by talking to the measureware daemon.

      Glance is a real-time tool with both a terminal and an X-GUI version. It allows you to view all of the statistics on a global level or drill down to an individual thread. It gives you network I/O, disk, general I/O, memory, system-calls, CPU loads, context switches, etc. All-in-all there must be hundreds of statitics available. There is also an integrated help system which explains how all of the statistics are computed (which helps compair statitics whith other tools on other platforms).

      PerfView gives you the same statitics over time (days, weeks, months or years, depending on how much data you choose to keep) averaged into 5 minute samples. You can generate ad-hoc graphs, or save them for later use. You can also look at multiple systems on your network. For instance you could compare CPU, Network and Disk I/O statistics for a cluster of web servers (what I used to do) to make sure they are evenly loaded. You can also look at filesystem utilization over time. This is particularly helpful for figuring out how fast a filesystem is growing before it is too late (like /tmp or a web-log directory).

      Finally there is OpenView Reporting server which I have not used. This allows you to take the same data, and use "canned" and custom reports to automate reporting. I believe you could have it email or FTP you a CPU report for a collection of machines every morning, etc. I believe various other parts of OpenView can use the measureware data as well.

      The only down side to the whole thing is price. HP MeasureWare and it's associated OpenView products are outrageously expensive. I was really trying to sell the last company I worked for on MeasureWare, but they just could afford it. I believe just the agent (daemon) is a few thousand dollars. PerfView was in the ball-park of 10 - 15 thousand and Glance was a couple grand.

      I hope Sun's new utility will push down the pricetag, but knowing HP, I'm not counting on it.

      P.S.

      I'm not an HP employee, I'm a Unix Sysadmin, who has worked with HPUX, Linux, NetBSD, Mac OS X, Solaris and other OSs.

  32. Re:Lots of users == good. by afidel · · Score: 4, Informative

    Well you CAN get Solaris 10 preview from the Express program. It's free for non-commercial use and available for x86. Downloads available here. Check out the hardware compatibility list for x86 here (don't see one specifically for Solaris 10 Express, probably the Solaris 9 list applies).

    --
    There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
  33. Linux Kernel Instrumentation by DaGoodBoy · · Score: 4, Informative

    The Kernel Instrumentation Process has already started creating the foundation for a kind of dtrace functionality. The profiling mechanisms appear to echo some of what Sun has done to Solaris to enable this kind of process-less profiling. Hope someone is still pushing this along...

    DaGoodBoy

    --
    My God! It's full of Voids!
    1. Re:Linux Kernel Instrumentation by Anonymous Coward · · Score: 2, Informative
      Dtrace has significant advantages over kerninst
      1. It is 'safe', that is it is difficult if not impossible for a probe to break anything in a live system.
      2. It allows arbitrary aggregation of data
      3. It allows arbitrary predicates
      4. It allows arbitrary actions
    2. Re:Linux Kernel Instrumentation by elmegil · · Score: 1

      And it's implemented.

      --
      7 November 2006: The day Americans realized corruption and incompetence weren't addressing 11 September 2001
  34. Not really by Halo- · · Score: 4, Informative
    There are three issues with profiled libraries:
    1. The execution path is different between the profiled and non-profiled library, so you can not be 100% assured you are getting the same result.
    2. To engage a profiled library you have to stop and start the target process, and often play all sorts of pre-load and path games.
    3. Looking into the kernel requires even more invasion, and can mean rebooting the server.
    Basically, the idea here is that you don't have to use a possibly "different" library, and you can instrument code on the fly (without stopping the process or the box). This is really nice for production boxes where any interruption or change of software is a really big deal.
  35. Can someone summarise what Dtrace is? by AliasTheRoot · · Score: 1

    My head explodes when I read marketing speak, what does it do? what is it?

    (any answers along the lines of it DTRACES your OS etc will make me come to your mothers house and explode my head all over her prize dinnerware set)

  36. Dtrace will put me out of a job! by Ayanami+Rei · · Score: 0
    --
    THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
  37. Well... by Ayanami+Rei · · Score: 1

    it gives you access to all Solaris' performance counters, system call tracing, etc. and gives you a command interface where you write simple scripts that easily manipulate and present that data in a way which is useful to your performance tuning or debugging purposes.

    You could derive work-alike programs for top, sar, ps, etc. etc. using simple dtrace scripts.

    And a lot of the statistics would be normally nearly impossible to get without debugging the kernel or poking around in /dev/kmem.

    By putting it all in one place, and putting a nice interface on it, all of that is in the past.

    --
    THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
    1. Re:Well... by AliasTheRoot · · Score: 1

      So basically it makes Solaris Sysadmins obsolete?

      Anyone know of a BSD gig going?

    2. Re:Well... by Anonymous Coward · · Score: 0

      It'll make Solaris Sysadmins obsolete - and Solaris Sysadmins who know DTrace worth their weight in gold. After all, someone has to use the tool in the first place ;-).

  38. Exists already in Linux kernel by johnhennessy · · Score: 1

    This already exists in the linux 2.6 kernel (I think it may be back ported to the 2.4 as well).

    Its called OProfile.

    While not identical to DTrace it lets you perform the same task.

    And another advantage of open source - if you do find a resource hog, you can go in and "make it right".

    Heck, for really silly bottlenecks you could fix it in hours/days rather than months.

    --
    [ Monday is a terrible way to spend one seventh of your life. ]
    1. Re:Exists already in Linux kernel by movement · · Score: 5, Informative

      I wrote OProfile, and I currently use DTrace daily, and I can assure you that you are wrong when you claim they do the same thing.

      OProfile is useful for measuring system-wide resource consumers (for example, you can see what pieces of code are causing cache misses in the kernel when your apache process is in the kernel etc, or which user processes take up the most CPU time).

      DTrace can also do something similar (though it needs a little more work yet). But DTrace does a LOT more than this. Imagine a system-wide (kernel, binaries, libraries) 'strace', where you can trivially choose what to print out, and what parts to strace, and under what circumstances. DTrace does even more than that.

      OProfile can't tell you exactly why your system call is returning EINVAL. OProfile can't tell you why your application is causing cross-calls. OProfile can't tell you what processes are writing to what files, in real time. OProfile can't debug race conditions.

      OProfile is a profiler: it does its job and nothing more. DTrace is, essentially, an instrumentation suite; one of its abilities is to function as a simple profiler.

      You won't really get a notion of why DTrace is so useful until you try it.

      --
      -- Remove the trailing '\0' to email me.
    2. Re:Exists already in Linux kernel by 110010001000 · · Score: 1, Funny

      "OProfile can't tell you exactly why your system call is returning EINVAL. OProfile can't tell you why your application is causing cross-calls. OProfile can't tell you what processes are writing to what files, in real time. OProfile can't debug race conditions."

      Well then what are you posting on Slashdot for??? Get back to work! :-) Just kidding.

    3. Re:Exists already in Linux kernel by Anonymous Coward · · Score: 0

      wow, now I feel like buying a new Sun box for my next development server instead of an AMD Opteron box. thanks for the info.

    4. Re:Exists already in Linux kernel by MROD · · Score: 1

      Of course, you can merely get Solaris 10 x86 when it comes out then you can use the Opteron box if you want to.

      --

      Agrajag: "Oh no, not again!"
  39. Linux is not Solaris. by ehack · · Score: 5, Interesting

    I don't think Sun's open sourcing the thing will help Linux users much anyway, the kernel design is probably very different, with more evolved semantics. Which means Linux developers will hae to do the work themselves if they want their admins to have these facilites.

    Note that this is finally a tool to resolve finger-pointing *it's your app that's slow - no it's the kernel etc* . I guess IBM will find it advantageous to create a similar Linux tool for their systems, as they bill Linux as an application serving environment. So we should see it soon on our own boxen.

    The sad thing is this proves once again that closed source companies retain an edge at innovating.

    --
    This is not a signature.
    1. Re:Linux is not Solaris. by Macka · · Score: 1
      The sad thing is this proves once again that closed source companies retain an edge at innovating.
      I think you're missing the point. It's competition that drives innovation. Sun are having a hard time keeping up with performance of their rivals in no small part due to the fact their cpu architecture is falling behind. In order to remain competitive they need to take advantage of every opportunity to squeeze more performance from their systems and apps. If that urgency didn't exist, would they have felt the need to develop Dtrace? Probably not.

      A big part of that competitive pressure these days comes from Open Source computing, because the advance of Open Source in areas of OS/kernel/app development has been so rapid and relentless. Closed source companies have to innovate, and hard, or they'll eventually get over run by the Open Source competition.

      Many of them are doing just that. Sun seems to be one of the closed source kernel/OS leaders. Others like Apple are in a similar position on the desktop. Microsoft conversely feel the competition less than the others because of their monopoly status, and you can see how slow and sluggish they are to genuinely innovate. e.g. How many years has it been now since they released a version of IE with any innovative new features in it? All they seem to do is plug holes in the crumbling dam of an excuse for a security model.
    2. Re:Linux is not Solaris. by /ASCII · · Score: 1

      The sad thing is this proves once again that closed source companies retain an edge at innovating.

      I disagree. Dtrace is the next logical step in performance monitoring. It is ahead of previous tools like gprof, gcov, Vampir, etc, and as such it represents real inovation. But is is not a revolution.

      I can think of dozens of Open Source programs which have also taken the next logical step in some field, and are in my opinion just as innovative. These programs include Screen, Valgrind, Beagle and BitTorrent.

      Longhorn seems to contain a lot of innovative technology. It remains to be seen if Longhorn will ship in a form even remotely simmilar to what is promised, or if most of the features die on the way.

      Regardless, a lot of people are already trying to create Open source solutions to the same problems solved by Longhorn. The Mozilla foundation is trying to create a new rich format for the next generation Web interface, ReiserFS aims to support database like querying in the filesystem, the freedesktop X-server promises a sane, modern windowing environment, etc, etc.

      Some might say that these people are copying Microsoft inventions, but I disagree. I would say that these people are innovating, most of those ideas are rather obvious and have been around for decades, the trick is to implement them and integrate them into the rest of the system.

      --
      Try out fish, the friendly interactive shell.
  40. Other tools that do this. by welloy · · Score: 5, Interesting
    "For example, there is no tool anywhere that allows for arbitrary dynamic instrumentation of a production operating system kernel."

    actually that's not true. Kerninst does exactly that.

    Kerninst is a framework for dynamically splicing code into a running kernel, almost anywhere, anytime. Code can be removed and changed at will. Kerninst works on standard (unmodified) Solaris and Linux kernels, *no* kernel re-compilation is necessary.

    1. Re:Other tools that do this. by Dug · · Score: 5, Informative
      Check out the Dtrace Usenix paper "Dynamic Instrumentation of Production Systems".

      This paragraph is copied from the Related Work section.

      Kerninst is a dynamic instrumentation framework that is designed for use on commodity operating system kernels[13]. Kerninst achieves zero probe effect when disabled, and allows instrumentation of virtually any text in the kernel. However, Kerninst is highly aggressive in its instrumentation; users can erroneously induce a fatal error by accidentally instrumenting routines that are not actually safe to instrument.3 Kerninst allows for some coalesence of data, but data may not be aggregated based on arbitrary tuples. Kerninst has some predicate support, but it does not allow for arbitrary predicates and has no support for arbitrary actions.

    2. Re:Other tools that do this. by Anonymous Coward · · Score: 0

      Kerninst was really cool, but in essence a hack. What I mean is that it may have been useful on a test system, but I would not trust it on a live production system. With Kerninst you were basically patching in binary code into a running kernel with little synchronization. And on Sparc at least, you had to disable kernel module unloading because the FINI's for kernel modules scattered around the address space were used as 'jump islands' to get around the kernel address space to the hot-patched code. To be fair I do not remember exactly what the details were about synchronization but I do know that it paniced the node I saw it running on with some frequency.

    3. Re:Other tools that do this. by Anonymous Coward · · Score: 1, Informative
      actually that's not true. Kerninst does exactly that.

      Kerninst is a framework for dynamically splicing code into a running kernel, almost anywhere, anytime. Code can be removed and changed at will.

      Let me rephrase what you are responding to in the hope that this will make things a little clearer:

      there is no tool anywhere that allows for arbitrary dynamic instrumentation of a **production** operating system kernel

      The point here is that, yes, you could go inserting arbitrary code into any random point of a kernel on your mission critical big server whose downtime causes real dollar losses every minute it's down, but is this a smart thing to be doing? If you read the Usenix paper on it, you'll see that one of the design requirements from the beginning for dtrace is for it to be totally safe to run this tool on a machine. In theory, at least, you cannot do anything with dtrace that will crash your machine.

  41. Re:Lots of users == good. by mollymoo · · Score: 1
    I think I got what you meant, but I didn't necessarily agree with it.

    I think your point was that wide availability of a tool will increase the number of people using it and thereby increase sales of (in this case) Sun products. While that is true to an extent I think you're hoping for a bit much for Sun to port it to Linux. Sun are a business. They have an OS and boxen to sell. They are generally a community-friendly company, sure, but if dtrace is as useful as they say then I can see them wanting to keep it proprietary and use it to draw people in to buy their OS and their boxen.

    I'm not even sure how well other OSes could use dtrace, at least without a fair bit of work. It sounds to me like it's pretty tightly integrated with the OS (for obvious reasons).

    And anyway, you can use it. You can download Solaris for x86 for free and get used to the tools that way. So basically they are doing what you want, but with their entire OS instead of just this one tool.

    I'm so nice I'll even show you where you can get it. Right here.

    --
    Chernobyl 'not a wildlife haven' - BBC News
  42. Favorite quote by kill-hup · · Score: 5, Funny
    DTrace's inventors say admins need "to have a good relationship with their brains" to use the software best.

    I'll have to start using that in my conversations about lusers. "I wouldn't call him stupid; let's just say he doesn't have a good relationship with his brain..." :)

    --
    Sinepaw.org: Grape Winos
  43. Re:Isn't this a profiler for non-profiled binaries by Anonymous Coward · · Score: 2, Informative

    Aside from the fact that this tool works with non-profiled binaries, I fail to see the revolutionary aspect.

    try reading some of the related links (like the bigadmin dtrace site, the usenix paper, the com.unix.solaris post, or the sys admin magazine article) and you'll quickly realize why this is a revolutionary tool.

    the profiling facility is only one of the information providers that dtrace has.

    other providers include:
    - a function provider (allows you to monitor calls into any kernel function entry or exit point )
    - a system call provider
    - io provider
    - process scheduler provider
    - process provider (allows you to place probes in userland processes.)
    - the list goes on

    these providers taken together allow you to place probe points all over the kernel and userland processes where each probe point can return whatever information you want. (ie, you can get function argument/return values, you can access globals, you can troll kernel data structures, you can collect you own statistics, etc.)

    that's a lot of potential information, so dtrace provides you with an easy programmatic way to drop data your not interested in. how about only collecting data when a particular function returns an error after being called from a particular process with a certain parameter set. easily done in dtrace.

    then your simple dtrace script can correlate information from all these providers and display it clearly. (with a language which was designed to make simple statistical analysis easy via built in functions for aggregations, averages, etc.)

    and of course the best part is that you can do all of this on a production system, running production software, and under normal load.

    obvious disclaimer: i work for sun.

  44. Sun comming arround? by argoff · · Score: 3, Insightful

    There's an old saying, "you are what you hold yourself accountable to".

    In that sense, Sun is no Microsoft. They are a hardware company that provides services, plain and simple. The only reason why they are cold to Linux is because it pits Sun servers directly against x86 commodity PC's - otherwise they are all for it.

    Once the dust settles on their bread and butter revenue stream, you can better believe that they will be open-source all the way. But right now they need to force some differentation with their hardware because in most cases they simply can't compete against an x86 farm in the server space. All the rest of the BS about new accomplishment is just propaganda, I would ignore it.

    1. Re:Sun comming arround? by christophersaul · · Score: 1

      Apart from the fact that they sell x86 and Opteron hardware and would happily have you running Solaris x86, which has DTrace as well, or Linux, bought from Sun, on that server farm, with Sun's 24x7 support for the whole solution. That server farm's probably connecting to a Sparc Oracle database anyway, so you can get the whole lot from Sun.

      Find out about what DTrace is, can do and what it means before simply referring to it as 'BS'.

    2. Re:Sun comming arround? by argoff · · Score: 1

      I read about dtrace in detail, it is quite cool. Of course they are releasing it, because that kind of tool is usefull far more often on high end high cost computers, but it is still cool. When I was refering to BS, I was refering to all the other propaganda coming from Sun. Like ....

      No .... they would not have you hapilly running Solaris X86 - they only hapially do that when it becomes obvious that they cant push a bunch of overpriced SPARC servers on you. I guess they figure it's better to have you as a customer on an x86 then to have no customer at all.

      Once again, you are what you hold yourself accountable to. In most situations, Sun simply can't compete against the x86 head to head. I know that offends many a Sun engineer, well sorry - it's not personal - I didn't create that reality.

    3. Re:Sun comming arround? by Anonymous Coward · · Score: 0

      What's so overpriced about paying $50k for an 8-way server?

  45. Ahh.. by [tsa] · · Score: 1

    ..finally Sun gets innovative again. I guess Solaris
    10 will be the best release since 5.7..

  46. So it's the SE toolkit by astrojetsonjr · · Score: 1
    Fine, so these blokes have written a wrapper report for the SE toolkit, a BOFH "instrument" for years. Nice job guys, let me know how it works out for you.

    Sigh, so now the PHB will be running this and telling us what to fix.

    1. Re:So it's the SE toolkit by Keith+Maniac · · Score: 2, Informative


      DTrace is not the SE toolkit. The SE toolkit used a similar method (little langugae scripts) to collect data that was made public by the kernel.

      DTrace takes advantage of monitoring hooks placed throughout the kernel and in every system call, and it does things that SE never could.

      Furthermore, one look at the DTrace manual and your boss will shit his pants. This is not PHB material by any means.

      Chapter 1: The D prograsmming Language...

    2. Re:So it's the SE toolkit by Anonymous Coward · · Score: 0
      Chapter 1: The D prograsmming Language...

      Thank you for adding a new word to the geek lexicon, meaning an unnatural state of excitement over a programming or scripting language. "Man, have you tried D yet!? I just about had a prograsm!!"

    3. Re:So it's the SE toolkit by Keith+Maniac · · Score: 1

      That's the sort of innovation that comes from typing quickly because you're late for work, but in the middle of a post.

      I happily release 'progasmming' and all the other typos in that post into the public domain for unrestricted use.

  47. Sounds very cool... by teamhasnoi · · Score: 4, Funny
    I wonder if they have attempted to run Dtrace on Dtrace? Wouldn't that allow you to make Dtrace infinitely fast? Perhaps become its own energy source?

    I know I would pay big dollars, say like $28, for a box that ran infinitely fast.

    This could be the diet pill for software bloat that we've all been looking for.

    1. Re:Sounds very cool... by teamhasnoi · · Score: 1
      For the record - I RTFA, and although I am not a dev, think I have a handle on what Dtrace does.

      Would this *not* be an antidote to code bloat? I am rather puzzled why my parent post was slapped with -1 troll. Is bloat dear to the moderator? *hint hint*

  48. YOU FAIL IT by Anonymous Coward · · Score: 0

    You have failed to define statement a. Try again.

  49. Tunnel Vision Sysadmins by AnomalyConcept · · Score: 1

    (slightly offtopic, but is pertinent to this thread) Wasn't diversifying hardware and software one of the ways that Akamai defended itself against the massive DDoS attack?

  50. Solaris is still around? by plutoiddiamonds · · Score: 0, Flamebait

    Why don't they replace solaris with linux? They should start becoming more mainstream.

  51. Re:And what is DTrace - Shark I presume by sjf · · Score: 2, Informative

    Sounds just like Shark for Mac OS X.
    If DTrace is better than Shark, it must be an awesome tool. Shark does everything this article describes and far, far more:

    http://developer.apple.com/tools/shark_optimize. ht ml

    The 4.0 version of Shark shown at WWDC was even better. In many cases, Shark can actually give you specific directions on how to optimize your code: both at the source and instruction level.

  52. Sun has a long history of doing both at same time. by Ungrounded+Lightning · · Score: 1

    so, we are assured of a and !a ???

    Sun has a long history of vacilating on whether and how to open their products. Hardware interfaces, SPARC-based hardware cloning, various software packages, and so on. Often this means half-opening, licensing and failing to renew, and so on.

    It seems to be different every time. But the common thread is that they never commit. (I think the closest they came is Open Office where at least we have a fork.)

    It was due to their refusal to open things sufficiently for me to hack my Sun hardware and software that I switched to Linux a few years back (and swore off anything but open source for my home machines), rather than trying to upgrade my Sun systems for the coming millenium.

    Came Y2K not a lot of stuff broke. But I shut down the last two anyhow. Too late, guys!

    (And too late for me too, apparently. Things have been so busy in the day job that I haven't had a chance to actually DO any hacking at home - except on the houses themselves. B-( )

    --
    Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
  53. I'm confused... by Trevin · · Score: 1

    ... by the article's use of the word 'instrument', as in "Users have more places to instrument a box and an easier way to do so.". I had never heard it used as a verb before. So I looked it up in the dictionary:

    tr.v. instrumented, instrumenting, instruments (-mnt)
    1. To provide or equip with instruments.
    2. Music. To compose or arrange for performance.
    3. To address a legal document to.

    Can somebody tell me which of these definitions is meant in the article?

    1. Re:I'm confused... by LostCauz · · Score: 0

      my guess would be #1, as in the ability to place "instruments" in more places to monitor what is going on.

    2. Re:I'm confused... by mistshadow · · Score: 1
      Try a computing dictionary:
      From The Free On-line Dictionary of Computing (15Feb98) [foldoc]:

      instrument

      <programming> To install devices or instructions into hardware
      or software to monitor the operation of a system or component.

      (22 May 1996)
    3. Re:I'm confused... by Anonymous Coward · · Score: 0

      Can somebody tell me which of these definitions is meant in the article?

      None of those definitions. In this usage it is a technical term and not covered in your dictionary. Try looking up pop and see if it talks about stacks.

  54. Re:Lots of users == good. by Anonymous Coward · · Score: 0

    "You can download Solaris for x86 for free"

    No you cannot. You can download it for no money but it certainly isn't free. Please stop confusing "free as in speech" and "free as in beer", it only confuses the newcomers to Slashdot.

  55. Insure/Ensure by Anonymous Coward · · Score: 0
    Although I prefer the use of "ensure" in this context, the dictionary says they're interchangeable for this usage in the usage note on this page:
    Although ensure and insure are generally interchangeable, only insure is now widely used in American English in the commercial sense of "to guarantee persons or property against risk."
  56. PTrace == DTrace ? by Anonymous Coward · · Score: 0

    After RTFA DTrace remind me of PTrace(3) (along with strace). I think ptrace can do the same thing, and more..

  57. Obligatory Simpsons quote by bastard42 · · Score: 1

    Alright brain. You don't like me and I don't like you, but let's just do this and I can get back to killing you with beer.
    Homer

    I try to use beer as a motivator for my brain (Do this and you won't have to think for a little while). I don't think it works.
  58. Re:And what is DTrace - Shark I presume by kscguru · · Score: 1
    I'll be blunt: Shark and DTrace are two entirely different animals.

    Looking at that Apple blurb (and reports I heard from WWDC attendees), Shark is a performance tweaking tool - a profiler. If you had ever bought the heavy-duty Sun compilers, you'd have a functionally similar set of tools (collect, analyzer, and cc with various flags that generates annotated source). Admittedly not nearly as polished as Apple's tools (and not as cheap), but Shark is also several years newer - and having done exactly the optimization that article describes on my own code with the Sun tools, I assure you they have been around for quite a while.

    However, DTrace is NOT an application profiler - though it can be used for that. And claiming that it is a profiler would be like claiming C is just an extension to assembly language. It could be used as such, but you'd be ignoring all the power of the tool.

    DTrace is far more powerful: it is a whole system profiling tool. It can give debugging data, performance data, or whatever, for any process, thread, driver, kernel module, or other system component. It includes a moderately complex programming language - YOU get to write code that runs in REAL TIME within any code path anywhere in the system. And, you get zero performance loss in any part of the system that isn't actively being traced.

    The DTrace manual is extremely informative. Of course, actually trying it on Solaris 10 Beta is even more so :-)

    --

    A witty [sig] proves nothing. --Voltaire

  59. Dubya by Rufus88 · · Score: 0, Offtopic

    "I wouldn't call him stupid; let's just say he doesn't have a good relationship with his brain..."

    Nah, George Bush and Karl Rove are getting along just fine.

  60. Your quote is too short by Anonymous Coward · · Score: 0

    Weren't the Phiilosophers in this story eventually placated by the fact that they would be neeeded to explain various possible reasons for the machine's answer?

    1. Re:Your quote is too short by tgrigsby · · Score: 1

      Weren't the Phiilosophers in this story eventually placated by the fact that they would be neeeded to explain various possible reasons for the machine's answer?
      Wait a minute: they were told they'd go from being R&D to being tech support? And they were ok with this? Yeah, that's fiction alright.

      I went from 3.5 yrs of pure tech support to R&D and never looked back. Sorry, scuds, but if there isn't at least a *little* creativity in the job, I go nuts....

      I mean, look at me now......

      --
      *** *** You're just jealous 'cause the voices talk to me... ***
  61. ^^ not || by Psymunn · · Score: 1

    In english, or is generally used to specify one or the other, but not both. So, they where really saying 'a xor !a'. Granted, this still evaluates to true.
    And it still dodging around the issue. I do enjoy how project looking glass (cool but fairly useless) was used as an example of the trend of Sun to go open source. I guess, as people speculated, Sun decided to dig through their stuff, find something 'sorta cool' and throw it to the public saying 'see, we're trying.

    --
    The Neo-Bohemian Techno-Socialist
  62. interesting experience (perhaps kind of OT) by kardar · · Score: 1

    I have this somewhat funky, older computer - it's a somewhat (1.5 Ghz) speedy Athlon, but the thing is it has this somewhat older Seagate 10K RPM SCSI hard drive in it. I kind of pieced this computer together from parts here, and parts there.

    So of course, I had to install Gentoo on it. I did. It took a while. I burned out a stick of RAM. I am afraid of Gentoo. But anyway, I got it up and running, but I had to disable tagged queueing on the disk to keep it from freezing up all the time. So here I am, with a functional Gentoo machine, and this 10,000 RPM disk is very, very loud. You can hear it in the next room when it does an updatedb or anything like that. With the ext3 filesystem, I do believe that it was doing what you call "thrashing". Just really loud, scary type of many tiny ball bearings in a coffee can type noise. Well, I had had enough, and Gentoo really creeped me out. I didn't like the idea of a journaling filesystem that fscks every now and then and you never know what the fsck is going to find. The first fsck it did (after some 20+ reboots over the course of a month or two) required a reboot because of some inconsistency it found. Just not cool at all - maybe I am paranoid, but I would rather have the thing fsck right away - and I know you can set it that way, or set it not to fsck at all, but in any case - the real drag was how loud the disk was, and that it appeared to be thrashing when certain things were going on - make installs, file extracts, updatedbs. I didn't like it. It creeped me out.

    So I installed FreeBSD 4.10. I love FreeBSD. And you know something, the disk is really, really quiet now. I went back and re-enabled tagged queueing. No problems. Same disk, same controller, different OS. The disk sort of sounds like a welder now (like you are welding something). It's not loud, it doesn't thrash, and the power got cut once, but it fsck'd right away and you know you're good to go immediately, not waiting for the next 28th reboot to find out there is something horribly wrong with your filesystem.

    So it's really amazing to me, how much quieter the disk is with FreeBSD. I know it's a loud disk, it's kind of a funky computer. It even has a funky heatsink. But the difference is like night and day.

    So in this case, switching from ext3 to ufs solved a problem that I had. The disk I/O is actually a tad slower with ufs, it appears, according to bonnie++, which I did on both OS's... but the disk is so much quieter that I can hardly believe it. I would have never imagined it.

    So sometimes, it may not be so much that you know what is wrong, or where the problem is coming from, but it's your ability to do something about it. In my case, I chose ufs, because I didn't want the disk to thrash, even though I was getting just slightly over 1MB/s greater disk I/O with ext3. This is not something you can just do overnight, it required a complete OS reinstall.

    Right now, though, it is just a really wonderful computer to use - my last uptime was 10 days, 10 hours, and 10 minutes and 39 seconds. (storm was coming in and I don't have a UPS). Wonderful, stable, smooth-running, and _somewhat_ quiet system.

    But seriously, I still haven't figured out what it is about Gentoo that creeps me out. I wonder if Dtrace could help me out with that one.

    Dtrace does sound cool though, I think it can be very useful, provided that you can make the changes that you need to make without reinventing the wheel, or refactoring your entire program. Maybe there would be some way that Sun could sell it to people, or port it to other OS's and let people buy binary versions of it for a nominal fee?

    1. Re:interesting experience (perhaps kind of OT) by Crackez · · Score: 1

      Just a wild a$$ guess, but i read this in a book about the BeOS FS. Basically Berkley FFS, you know it as UFS as a FreeBSD user, introduced the concept of cylinder groups to the file system. Since you didn't really say much about your drive other than it was a 10k Seagate drive, I have to make a few guesses. I have similar drives in a drive cage, attached to my Ultra 5 (2x 9GB Seagate Barracudas) and again, I am just guessing, but, these drives probably have multiple platters. You're average IDE drive only has one platter for cost effectiveness... So whats my point? Ext3 (ext2 really) ditched the cylinder groups idea since alot of the newer IDE tech has similar technology built into the controllers to optimize throughput, or so it says in the book. Soooo, using cyilder groups probably tends to give a nicer phyical layout on the disk then assuming the hardware will do it for you like ext3 does.

      Maybe thats why your disk appears to thrash more with ext3 than with FFS... Especially during the disk intensive tasks you mentioned.

      ----
      Anyone wanna hire a just graduated college guy, just outta school?

    2. Re:interesting experience (perhaps kind of OT) by Anonymous Coward · · Score: 0

      So, of course, somewhat boring. So. Somewhat. So, sometimes somewhat.

      But seriously, somewhat Milton, where's your stapler?

  63. Nice automated Gui? by Fiz+Ocelot · · Score: 1
    So lets say I wanted to make a dtrace function that could do something like monitor for a high amount of cross calls. Upon noticing such an occurance it will then gather information on it and create a nice breakdown for me in an easy to see interface.

    Or I guess I'm asking, how complex can these functions get? Would it be feasible to tie it in with other things so I could do my own interface to it? Sounds like fun if I can...

    1. Re:Nice automated Gui? by Keith+Maniac · · Score: 1

      No GUI. (yet)

      These functions can get as complex as you'd like, since DTrace is driven by the 'D' programming language.

      Here's a few good examples showing the use of speculations (only tell me when a syscall *doesn't* work, not when it does) and aggregations (summarize the data, so you don't have to).

      http://blogs.sun.com/roller/page/ahl/20040701#dtra ce_for_developers

      This sort of built in functionality is what separate DTrace from tools like truss, not to mention the huge number of OS hooks those other tools can't touch.

  64. Re:Lots of users == good. by slickepott · · Score: 1

    Yeah.. stop confusing that.. That sentence says it can be downloaded at no cost and not that it's free to use any way you want.

  65. Re:niGg4 by Anonymous Coward · · Score: 0

    Fuck the what?

  66. Re:Hmmm by Bogue · · Score: 1

    Give system catchy name, vaguely reminiscent of Eminem's rap group

    I'm sure they did this to appeal to the disproportionately-large numbers of African-Americans in the field of computer science.


    The majority of Eminem's fan base is white.

  67. Dtrace == New Shimmer??? by HydeMan · · Score: 1

    Its a dessert topping, AND a floor wax.

  68. A competing tool for Linux by catalina68 · · Score: 3, Interesting
    For example, there is no tool anywhere that allows for arbitrary dynamic instrumentation of a production operating system kernel.
    I guess these guys haven't heard about SpyKer from LynuxWorks. http://www.lynuxworks.com/products/spyker/spyker03 .php3
  69. Code Review by ansible · · Score: 2, Insightful

    I was reading the examples of using DTrace to spot performance issues.

    It seems to be that most of them could have been caught with code reviews.

    Perhaps with tools like DTrace, you can spot performance issues more quickly. But I believe a good code inspection regime could catch these problems and more.

    1. Re:Code Review by Anonymous Coward · · Score: 0

      > It seems to be that most of them could have been caught with code reviews.

      Who would then prove it ... by firing up DTrace. I don't know what kind of code reviews you do, but I never did mine down to CPU instructions...

    2. Re:Code Review by k8to · · Score: 1

      Code reviews are good tools. DTrace is apparently a good tool. They are not redundant.

      I can't code review code that some other company wrote, but I can trace it to identify a problem I am having while using it. Heck, I may not even understand the entire design of some app I've been given, but I can use DTrace to try to figure out why its network response times seem so sluggish. It could be something about how our systems are set up, or it might become an obvious common performance issue. Not that you couldn't necessarily investigate some of these issues in other ways, but supposedly DTrace provides a more powerful, direct, and rapid way of doing said analysis.

      A second response I have is that many code reviews I've been part of have involved a lot of nudge nudge wink wink. If yours are thurough, efficient, and not onerous, count yourself truly blessed.

      --
      -josh
  70. Sun is finally catching up... by Anonymous Coward · · Score: 1

    ...with HP-UX, which has had MeasureWare and PerfTools for years. HP even makes a version available for Solaris (and has for some time), as well as AIX and Windows (and probably Linux by now, but I don't know for sure).

    I know this looks like a shameless plug, but I hate it when companies push some mundane thing as being "revolutionary" when it has basically been around for over a decade. Hey Sun, you know what be revolutionary? Open source Java. End of rant.

    1. Re:Sun is finally catching up... by Anonymous Coward · · Score: 0
      I hate it when companies push some mundane thing as being "revolutionary" when it has basically been around for over a decade.

      ...except that previous posts have pointed out that this really is something better than the tools you mentioned.

      What strikes me most about the commentary here is that the raves are coming from people who have actually used it, not from Sun (or not *only* from Sun; some people there seem justifiably proud of their work:-). The snarky comments are exclusively from people who haven't used DTrace ("gee, sounds like ____; what's new about that?"), and are being soundly rebutted by those who have.

  71. Re:Hmmm by Anonymous Coward · · Score: 0

    Way to miss the joke there, Poindexter.

  72. Does Dtrace benefit Scientific Computing? by Salis · · Score: 1

    So how exactly does Dtrace benefit application optimization? I code simulation programs to model physical systems. The applications can be CPU or I/O intensive and I'd like to know if Dtrace will help me optimize my code more than it already is.

    Can it suggest areas of vectorization, loop unrolling, parallelization, etc?

    Does it primarily give you stats on usage of functions, disk I/O, etc?

    The numerical methods I use are probabilistic, making them very hard to debug sometimes.
    (Imagine a bug occurring only 1e-4% of the time, although it will always occur before you get the answer you want. ;)

    Thanks for any info!

    --
    Favorite /. tagline: "On the eighth day, God created FORTRAN." And it was good.
    1. Re:Does Dtrace benefit Scientific Computing? by JonAnderson · · Score: 1

      You really need to read the documentation. It's not really for code profiling although you can achieve some of that (you can instrument every instruction in a user program if you wish). For a low frequency bug you can use speculative tracing which only commits the data when your specific event happens. i.e. you only get the data you are interested in.

  73. A guess by k8to · · Score: 1

    Perhaps it has to do with how much warning for bottleneck fillage is available in the real world situations encountered? It sounded to me like this was a sort of ASAP kind of fixing-it problem.

    Obviously your method would be great to pre-emptively schedule across all systems, but I suspect in a typical shop it might be one of those things the sysadmin means to do when the fires are all out. :-(

    --
    -josh
  74. Solarisx86 port? by Billly+Gates · · Score: 1

    I am begining to like professional unix more and Linux less.

    I am about to try out solarisx86 and my asus motherboard, nic, and video card all happened to be supported with Solaris 9 oddly enough.

    I am interested to give this thing a spin.

    1. Re:Solarisx86 port? by Tpenta · · Score: 1

      It is already available on Solaris Express (Solaris 10 preview) for x86 (in fact a lot of the develoment work started on x86).

      Tp.

    2. Re:Solarisx86 port? by int19 · · Score: 1

      I am begining to like professional unix more and Linux less.
      ...
      I am interested to give this thing a spin.

      Do it! :-)
      I tried it out for the first time four or five months ago, and just earlier today as a matter of fact I nuked my main home Linux install...

  75. VDtrace? by Anonymous Coward · · Score: 0

    Can I get Visual Dtrace for my NT boxen?

  76. Re:Lots of users == good. by Anonymous Coward · · Score: 0

    "That sentence says it can be downloaded at no cost and not that it's free to use any way you want."

    No it doesn't. It says that it's free when it isn't.

  77. Sun Reineventing the Wheel - again by msconception · · Score: 1
    Why does it always seem big news for Sun when they've done what is already available?

    And they don't even redo it well. Now, if dtrace were user/admin friendly, I could see why a new tool had to be reinvented. But dtrace is a step above writing a c program to use probe system calls - no imagination when it comes to what is useful to the user community.

    No doubt, dtrace is a powerful tool, but so is crash and adb and just as user-unfriendly. Sun should read the technology history books and listen to the user community before they tout capabilities that have existed in other OS's for decades.
    Instead they display their infancy to user-friendliness and mature OS capabilities. Kind of scary for an OS that's been around for 20+ years.

    dtrace appears to be another Sun internal science project to make it easier for Sun's own engineers to figure out what's going on in their OS. Their engineers, being young pups, have no clue that what they "invented" is not unique and pound their chests to display their inexperience in relating to the business world and user community.

    While IBM is touting technology that allows computers to read brainwaves, Sun is happy to give hackers yet another system monitoring tool (YASMT) and makes it front page news - no less. This is the type of tool IBM and others treat as ho-hum because they have much more relevant technologies to show off.

    1. Re:Sun Reineventing the Wheel - again by JonAnderson · · Score: 1

      This is was one of the most ignorant posts I have ever read on slashdot - and that is saying something!. Have you tried Dtrace? Can you tell me something that IBM has that even comes close? I would honestly be extremely interested. Try reading the USENIX paper: http://www.sun.com/bigadmin/content/dtrace/dtrace_ usenix.pdf Or see BMC's blog: http://blogs.sun.com/bmc He actually explains some of your FUD. Try and have an INFORMED opinion.

    2. Re:Sun Reineventing the Wheel - again by Anonymous Coward · · Score: 0

      Yes. I agree. Sun's powerful tools that lack user-friendliness does require full attacks to mask this weakness. In my opinion, lack of consistent GUI-based tools access is Sun achilles heel.

      Read the other posts on slashdot and you will see alternatives to dtrace on other platforms that have been around for a while.

      Sorry to touch so close to the nerve.

  78. MeasureWare? Sounds like Virtual Adrian by Anonymous Coward · · Score: 0

    Actually, MeasureWare sounds like the SE Toolkit (http://www.setoolkit.com/), formerly known as "Virtual Adrian", and now managed by Rich Petit. Also, Sun Management Center has some pretty good performance data aggregation capability.

    Also, is MeasureWare similar to SGI's Performance Copilot?

  79. Wow. Best reply ever! by Anonymous Coward · · Score: 0

    I'm glad to see authors of software posting.

  80. No Solaris 10 will be the best release since 2.5.1 by Anonymous Coward · · Score: 0

    Mark my word. Solaris 10 will be the biggest product innovation from Sun since the E10K, not just the biggest Solaris rev since Solaris 7.

  81. DTrace & Java by jo42 · · Score: 1

    Yes, nice, but will this tool help with improving Java app performance...?