Systemtap 1.0 has been released. There are a few features for this release, like experimental support for unprivileged users, cross-compiling for foreign architectures, matching C++ class and namespace scopes, reduced runtime memory consumption, but more importantly, this release means that Systemtap is finally considered stable and ready for user adoption.
And then we have ZFS wannabe: BTFS, and what more? I would love to see Linux inventing some new hot cool tech instead of just doing lesser copies from other OSes.
Sorry loser. Systemtap and BTFS is designed for the needs of *LINUX* users, something which ZFS and DTrace *ISN’T*
Now tell me which is superior?
Moron.
Edited 2009-09-24 09:52 UTC
Be nice, clei. Technically, he’s right. Linux specific or not, these are copies, albeit specialized.
I didnt get this. Linux users has no need of ZFS, but they have need of BTRFS? Linux users has no need of DTrace, but they have need of Systemtap? *perplex*
Look, I just want to say that I would like to Linux inventing some new hot stuff? Is that so bad? Dont you want that, too? Dont you agree that Linux copies from others?
Why do you care if Linux invents new stuff or not?
I think it would be cool if Linux that is the “best OS stealing market share from everyone – then it must be best” just proved it in some way. By developing some new hot tech that everyone needs to copy. A “killer feature”. ZFS has the protection against Silent Corruption and bit rot – which is ZFS killer feature. Never mind snapshots, high performance, etc. That is not interesting if your data is not guarded against bit rot and Silent Corruption. No other solution does this. And this is the only single reason to use ZFS. Even if ZFS was slow, had no functionality, had messy administration – people would still be drawn to ZFS. Because of Silent corruption. Read the spec of a new drive: “1 unrecoverable error in 10^15 bits read”. ZFS protects against that, no other filesystem does.
ZFS is a killer feature for Solaris. And also DTrace. DTrace IS are REAL killer. Even hotter than ZFS if you need to program or look into the inner workings of the system. DTrace is truly unique and never before has it been possible to do the things DTrace allows you.
So if Linux is the besto, please show it by being best. Be on the edge, tech wise. That would be cool.
Granted, something Linux is good at is the desktop experience. OpenSolaris has copied Linux desktop feeling by using Gnome. But where is the hot new tech? If Linux is the besto, then show it. Make me ditch OpenSolaris because Linux has cooler hotter tech.
Linux doesn’t need to be the best, just sufficiently good and ubiquitous. It’s a bit of same story as with MS-DOS, Windows, USB, PCI, VHS…
We would have a problem if Linux was not about to get the same features some time in the future. But we will, I don’t need them now, so I get by with Linux just fine. Hopefully Systemtap will be widely available soon as well… (I understand it still needs some patches for kernel? ptrace?).
First of all, let’s dispense with this silly idea that incorporating others’ proven features and ideas is bad. It’s not bad. It is the basis for how all of us move forward. (And yet this board sometimes seems rife with accusative the cry of “Copycat!”.)
Regarding ZFS… It implements some very cool and useful ideas. But IMO, a direct port would be a *terrible* fit even if the licensing allowed it. btrfs is still in development, and ZFS is in actual use. So we are behind on that. But I think it’s a much better situation than if we were trying to bolt ZFS onto Linux. So I guess I’d have to say “Thanks, Sun, for the inspiration!”.
I can’t really comment upon Systemtap. But it’s hard to imagine the kernel part porting very cleanly to such a very different kernel, even should the license allow it. So I would imagine that it would be an even worse fit than ZFS.
All in all, I guess I’m a Linux guy. But I was a Unix guy long before that. And I just don’t see a lot of value in all this rock throwing between camps.
Edited 2009-09-24 19:11 UTC
I’m say that I’m baffled by your post.
So, if Sun has dtrace, Linux must not have it’s own system performance gathering interface? Assuming that Linux cannot use ZFS due to licensing and design problems, should they stick to ext2? (ext3/4 has a journal, someone must have had it beforehand…)
Should Intel drop vtune just because Linux has oprofile? (or vice versa)
What’s next? should the GNU people stop working on GCC just because Watcom came first?
Where does it stop? Browsers? Editors?
Heck! Lets kill Linux! Minix came first!
… You do understand that how ridicules and childish your copycat claim sounds, right?
Edited 2009-09-24 23:25 UTC
Is it really? For years I have watched as the Linux users complained as Solaris users pointing to tools like DTrace and ZFS as reasons to select Solaris over Linux.
From the FCS (3/05) release of Solaris 10 which had Zones and Containers virtualization capabilities “out of the box”, it took RedHat and Novell 18 months to package virtualization capabilities in their respective products. And while RedHat and Novell have added some virtualization tools, there is no way to create a Container or a WPAR with either product. This is so they could stay competitive with Sun, IBM and HP.
The same goes with DTrace over Systemtap, why does it take Sun, IBM, HP or insert your vendor of choice here to produce some “cool tech” before someone thinks that “wow, maybe Linux should do that too”. I have said it before and I will say it again “Where’s the innovation at?” Because that is what it looks like when you start comparing feature sets of the various OS’s. While imitation is the sincerest form of flattery, at some point it stops becoming imitation and starts to look like copying what others are doing.
If Linux was “bleeding edge” it should be the other way around, Sun, IBM and HP would be trying to get the “cool tech” from Linux and that’s not the case. And while people here might not like Kebabbert’s opinion of the state of Linux feature development, what has he said that isn’t true?
… Don’t know, maybe Linux users actually need a copy-on-write FS? Performance gathering library? Maybe they actually need container based virtualization solution? Should RedHat simply give up on giving their customers what they need simply because it doesn’t measure up to your “innovation” standard?
Oh, lets not forget, if Sun starts following your “innovation” standard they should stop working on porting VirtualBox (which they own) to Solaris – after all, VMWare had it first!
Has it ever occurred to you that having a better, faster, more-resilient ZFS in the form of BTRFS * is -far- more important than your “innovation” factor? (In my short experience with ZFS, if something happens, good luck trying to salvage anything)
In end, I’d imagine that most users don’t really care about “I got there first” pissing contents – they (should) only care about what works, and works well. Who invented it first is irrelevant.
– Gilboa
My “innovation standard” as you put it is based on features that my customers and I want in an OS. If those features are not being developed then why should I use the OS? For an OS to be touted as “enterprise ready”, I expect enterprise features and not just a lot of pretty marketing speak. I don’t care about desktop eye candy and support for the latest USB wireless card since those features are shut off as part of our lockdown. This is the difference between desktop and enterprise computing, people like me are looking for the “cool tech” because it helps us give customers what they want and makes life easier for us.
Based on my experience I have had no issues with ZFS, sorry to hear about your problems.
Getting a bit defensive aren’t you?
Far from it.
I simply fail to understand what you want.
Maybe it’s a matter of a language barrier, but I’ve read your posts a couple of times and they register as being simple rants…. USB? What does USB has to do with ZFS/BTRFS and dtrace/systemtap?
– Gilboa
Since you are in your “on the attack” mode, I guess it’s fair to say that Solaris/OpenSolaris would be a complete nonstarter on my customers’ servers. For one thing, it couldn’t run their accounting or point of sale software. Oops…
And the problems and limitations roll on from there. The XDMCP servers need all the desktopish things that Linux offers. I guess maybe Solaris would be OK on the firewalls. But if I were going to move them from Linux, I’d be more inclined to go OpenBSD.
Glad you are happy with Solaris. But in my world, *Solaris is in a niche that doesn’t offer much of value for our business use.
I normally would just hold my tongue on this. But like I say, since you are on the attack… yet again…
Edited 2009-09-26 05:16 UTC
Well that is your opinion and you are entitled to it, just as I am entitled to mine. I don’t see the “innovation” in Linux and I have used it on and off for over 13 years. In many ways Linux either mimics or copies what the mainstream UNIX variants have done for years. And yes, I see Systemtap as a copy of DTrace.
This is wasting my time and getting very old. It is obvious I am not the only one who feels the way I do, otherwise Kebabbert’s comment would not be modded up to where it is. The difference is I am willing to say something and not just mod a comment up.
I’m really tired, having just finished driving 20 hours straight, or I’d post something eloquent, or dredge up a permalink. But you can probably google:
site:osnews.com sbergman “the i word”
or some such, to get my views on the relationship and relative value of selective incorporation of elsewhere developed features and the “i” word.
I’m not doubting Sun’s ability to “i”. But without agreeing with your claim than Linux doesn’t “i”… once the idea is out and proven… does it matter that much? Unless they are willing to attack OSS projects with patents. (Of course, there is a time element you could bring up, here, if you were so inclined.)
At any rate, I think what I am hearing is that you need the features of ZFS more than the application and driver compatibility, and mind-share of Linux. And I know that I most certainly need the application and driver compatibility, and mindshare of Linux more than I need ZFS. Which is fine. And it is far more than a different opinion. It is a different area of application.
We all tend to see our areas of application as most important. And there is nothing particularly wrong with that. It’s human nature. And face it. What is important to you is rightly more important to you the things which are equally important to me.
I think our common ground would be that both platforms benefit when one platform can fill an application area that the other might lose to a… well, let’s just say to a platform that we both would rather not see fill it.
Edited 2009-09-27 16:07 UTC