“Analysis Try to imagine a geeky version of famed salesman Ron Popeil. Keep Popeil’s exuberance, keep his pitchman savvy and keep his verbal overflow. Then erase his age, sturdy frame and Ronco Food Dehydrator and replace all this with a young, lanky kernel engineer hawking something called DTrace, and you have Bryan Cantrill.” Read the rest at TheRegister.
Sun delivers Unix shocker with DTrace
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
are these similar tools for linux/bsd? i am very impressed by the DTrace tool – another osnews-linked article linked to a brief walk-through… and it was very enlightening.
are these similar tools for linux/bsd?
Not even remotely.
The SUN engineers need focus. Dtrace is a swell product but they really need to address Solaris’s other defeciencies first. I highly doubt typical SA’s would use Dtrace for anything other than serious problems. Typical Sscenerio: YOu have been tasked to gather perfomance information for your servers and present them in a way that upper management can understand it i.e. a graph with large text and lots of pretty colors. HP-UX: fire up glance plus generate a graph from pre collected data and your done! Solaris: Write a 2 page Dtrace script export the data to some type of graphing program most likely a freesoftware utility(hope you can find binaries or you can compile it. Then hope for the best.
>are these similar tools for linux/bsd?
>
>Not even remotely.
>
Shrug. You mean for now.
In a couple of months you’ll be
seeing something similar or even
superior to them for both Linux and BSD
if the tools are really are what they are
cracked up to be and there is a large enough
group of people interested in the ideas
behind them.
Either way there’s nothing here to get
all that exicited about.
>are these similar tools for linux/bsd?
>
>Not even remotely.
>
I’m not sure of all the details, but IBM’s DProbes seem to be somewhat similar:
http://oss.software.ibm.com/developer/opensource/linux/projects/dpr…
Opersys has built a “dynamic tracing” toolkit, which uses dprobes:
http://www.opersys.com/ltt/index.html
are these similar tools for linux/bsd? i am very impressed by the DTrace tool – another osnews-linked article linked to a brief walk-through… and it was very enlightening.
Sun may put dtrace in the public domain at some point, so we may have these tools for bsd/linux at some point…hopefully.
Shrug. You mean for now.
Sort of. I’d expect a reasonable clone in the next three years.
In a couple of months you’ll be seeing something similar or even superior to them for both Linux and BSD
Don’t be silly. I am a BSD fan before anything, but this isn’t something that’s magically going to be cloned overnight by open source developers. Niether the BSDs nor Linux have the infrastructure to support anything like this ATM, and it’ll take time and a direction that all of these OSS projects currently lack.
if the tools are really are what they are cracked up to be and there is a large enough group of people interested in the ideas behind them.
Oh DTrace is, but you’re not going to see anything like it in any OSS OS anytime soon, least of all Linux.
Either way there’s nothing here to get all that exicited about.
Not on BSD or Linux that is, but Solaris users are getting something great.
> I’m not sure of all the details, but IBM’s DProbes seem to
> be somewhat similar:
DProbes and DTrace do have some attributes in common but
there are significant differences as well. It would be wise
to read the recent USENIX paper on DTrace that Bryan
co-authored as it dicusses some of these differences:
http://www.sun.com/bigadmin/content/dtrace/dtrace_usenix.pdf
And of course, the best way to understand what DTrace has
to offer is by downloading Solaris 10 and taking a look
at the engineer-written answerbook:
http://www.sun.com/bigadmin/content/dtrace/d10_latest.pdf
Oh DTrace is, but you’re not going to see anything like it in any OSS OS anytime soon, least of all Linux.
On what authority do you say that? There is talk that dtrace will be put in the public domain, which would make it immediately available to the OSS crowd. I agree that I don’t think a clone could be made overnight but it’s very possible that you will see dtrace in OSS.
Typical Sscenerio: YOu have been tasked to gather perfomance information for your servers and present them in a way that upper management can understand it i.e. a graph with large text and lots of pretty colors.
One word: Teamquest
And now, lets here about how would you done that on an AIX system please..
You only talked about HP-UX..
Thanks
PS: at least recognize what is _best_
Never had an oppourtunity to work with AIX in that capacity. Neve touched RS/6000 hardware until my current position so i am still learning. I thought i did recognize the best..Hewlett Packard..
The SUN engineers need focus. Dtrace is a swell product but they really need to address Solaris’s other defeciencies first. I highly doubt typical SA’s would use Dtrace for anything other than serious problems.
I’m really trying to bite my tongue here, but I find your post incredibly obnoxious — especially since you work for a competitor. I’ll ignore your claim that we “need focus” (which I find offensive, as it has taken an inhuman amount of focus to do what we have done in Solaris 10) and just focus on your “doubts” about the uses of DTrace. In our experience, typical SAs are using DTrace all the time for the most trivial of problems — and we’ve designed it to be that usable. In fact, many SAs are discovering that they now investigate minor problems that they used to just ignore — because DTrace has made it so much easier to investigate the problems. And look, don’t take my word for it; look at the customers weighing in for the Register article or look at the people weighing in on Slashdot.
This is what we want to see more of from Sun.
Less Redhat-bashing, less condescension, and more stuff that works to solve real problems.
I think DTrace sounds like excellent work from Sun, and while I can’t say I’ve ever needed to use any ‘dynamic instrumentation’ at the kernel level before, I can see how using such tools would allow you to tune things at a low level.
I wonder if DTrace is indirectly resposbile for some of the Java performance work in 1.5?
“This is what we want to see more of from Sun.”
Too bad it’s not very relevant until Sun open sources Solaris. It’s only relevant for those Sun customers who are going for Solaris 10 (which is exactly what Sun engineers are paid to do, obviously). The only reason to care about this more than caring about VMS getting a new color scheme is that Solaris is *potentially* open sourced one of these days.
I don’t see a real reason to be “enthusiastic” about this anymore than one is enthusiastic about some incremental improvement in any other proprietary product like MS Office. It’s just some stuff that someone out there hacked up that is not going to be available for most of us before Linux port/reimplementation.
Well, if Sun aren’t going to to be able to sell you any products, and all youre interested in getting from them is free stuff, I don’t think youre terribly relevant to them either.
As for me, i’d be quite happy to pay for Solaris if i needed DTrace. I think its great that Sun are producing tools that differentiate their product from Linux and other free OSes.
Basically, what you are suggesting is that no technology is ‘terribly relevant’ unless it is Open Source.
Thats an interesting position, but one that I cannot agree with, despite my widespread use and development of and for Linux.
That’s one of the reason for my scepticism about Sun in general. Schwarz has expressed that Open Source community in general is not relevant to Sun, because they are not their customers. That’s the kind of attitude that is going to get you new customers, or retain the old ones. The one that wants to sell you stuff is really in no position to be that arrogant – even Microsoft realizes this.
To a certain point it makes sense to ignore developers and focus on IT managers and people that buy your stuff directly. “Differentation” is exactly that – developers don’t want differentiation, they want a solid *commodity* foundation. It’s Sun’s responsibility to create “shareholder value” by differentiation, but that doesn’t mean we (developers, sysadmins, etc) should support them in their plans. If they are not willing to contribute to the common “intelligence pool” (OSS), they are only hoarding stuff for themselves and don’t deserve our support – of which even interest in their offerings is an example.
oh stop equating OSS with free software. redhat sells free software just fine. if OSS is irrelevant to sun why did they open source star office and talking so much about java and solaris adopting or not adopting such a model.
Oh stop getting your panties in a twist. The poster and the post you quoted made no comment about OSS the comment was directed at the author’s attitude not OSS.
Don’t just start flaming if you see the words “OSS” and “disagreement” in one post.
My point is simply that your statement that DTrace isn’t terribly relevant because it isn’t open source is stupid.
Theyre talking a lot about OSS because it makes sense in some situations to open source your software, and they are recognising this. I’d love to see it rolled into Linux etc. myself, but I don’t demand it, don’t expect them to do it unless *they* see a good reason to do so, and i’m just happy to see some news about Sun’s technical innovations rather than the usual name-calling.
I’m not saying its a bad idea to open source it, what i’m saying is that the licensing of a given piece of software is not the primary factor governing it’s ‘relevancy’.
I’m also saying that if you think a piece of software only becomes relevant when it it’s source is available for modification – I think that is wrong.
Most industries rely heavily on closed source/proprietary software – you can’t seriously be trying to say that products like Maya (visual effects), Pro/Engineer (MCAD), Microsoft Windows (buniess computing), Adobe Photoshop (Design), are ‘irrelevant’ because they don’t ship with source code?
Sure, they may not be relevant to you, but youre making a pretty big (and wrong) assumption when you speak for the rest of the world as to the relevancy of closed-source software.
I never said Open Source was irrelevant to Sun, never even suggested that.
I said *you* were irrelevant to Sun, how bout you clarify why you think you are?
“I’d love to see it rolled into Linux etc. myself, but I don’t demand it, don’t expect them to do it unless *they* see a good reason to do so, and i’m just happy to see some news about Sun’s technical innovations rather than the usual name-calling.”
I don’t demand it either – I just don’t think the technology is relevant until they do. The technology is relevant in the VMS color scheme way I mentioned earlier, i.e. some proprietary OS in its death throes has implemented something that helps its users. Hardly newsworthy.
“I’m also saying that if you think a piece of software only becomes relevant when it it’s source is available for modification – I think that is wrong. ”
It becomes relevant to a larger audience (non-customers) that way.
“Most industries rely heavily on closed source/proprietary software – you can’t seriously be trying to say that products like Maya (visual effects), Pro/Engineer (MCAD), Microsoft Windows (buniess computing), Adobe Photoshop (Design), are ‘irrelevant’ because they don’t ship with source code? ”
They are irrelevant to non-users. Of course they are relevant to the users, but for non-users, having public discussion about them is just free marketing for the producers, with no relevance to commnity-at-large.
“I said *you* were irrelevant to Sun, how bout you clarify why you think you are?”
I’m irrelevant for Sun because I don’t have purchasing power, and am not a Java developer. I don’t even know why I’m wasting my time on this thread…
I don’t demand it either – I just don’t think the technology is relevant until they do. The technology is relevant in the VMS color scheme way I mentioned earlier, i.e. some proprietary OS in its death throes has implemented something that helps its users. Hardly newsworthy.
Please prove to us how Solaris is in it’s death throes????
They are irrelevant to non-users. Of course they are relevant to the users, but for non-users, having public discussion about them is just free marketing for the producers, with no relevance to commnity-at-large.
I don’t use linux anymore. How is any linux discussion relevant to me??? you are claiming ubiquity with linux. That some how linux has far more users than any proprietary OS. You would be dead wrong with that claim.
I don’t even know why I’m wasting my time on this thread…
Becuase you are a linux zealot who can’t stay away from a Solaris thread.
Please prove to us how Solaris is in it’s death throes????
I thought it’s common knowledge. Read some articles, listen to industry analysts, whatever. Linux is already the “reference Unix” for which stuff is developed (developers love Linux, *because* it’s a commodity). Most positive articles about proprietary Unixen mention that “Unix is not dead yet”. Doesn’t that prove something?
you are claiming ubiquity with linux. That some how linux has far more users than any proprietary OS.
Linux is ubiquitous. Windows is ubiquitous. Solaris is not.
Becuase you are a linux zealot who can’t stay away from a Solaris thread.
Probably.