Let's take a closer look at OpenSolaris, particularly its use of ZFS, network problems that people have reported, the use of bash, and differences between OpenSolaris and Solaris and Solaris Express. Note: This is the latest article in our OSNews Article Contest.
To view parent comment, click here.
To read all comments associated with this story, please click here.
Member since:2005-07-08
There is a difference between shaman and segedunum, I have exchanged comments with shaman and he at least recognizes that some of Sun's technology is cool. Segedunum on the other hand has nothing nice to say about anything other than Linux. His other unique quality is his ability to pull stuff out of his posterior in order to support his position, an example of which is here:
http://osnews.com/thread?314698
And my response here:
http://osnews.com/thread?314699
His comments on my review aren't even worth responnding to, and his "problem" with running a python app on an UltraSPARC IV machine and it "magically" running better on a Linux machine is crap. Notice that he has yet to respond with any specific details as to why this was happening. If you have meaningful data that explains the behavior, show us. In their defense, neither shaman or segedunum got personal to my knowledge, like some people here.
Where I work, Solaris x86 got a bad rap because when they last used it in 1997 or 1998 it did not perform well. The poor performance I would attribute to not setting the IDE parameters correctly (if they used Solaris x86 on an IDE system). SCSI performance of Solaris x86 was good, based on my limited use. What bothered me is that nobody made any attempt to periodically check to see if anything improved. The shop I work in has been SPARC for years and will likely remain the same.
As far as compilers go, Sun wasn't the only company that did not ship a compiler as part of the OS. IBM doesn't ship one with AIX and one of my friends who used to manage IRIX machines told me that SGI wanted $20,000.00 for their C/C++ compiler. I have no experience with Sun's compiler, I have always used gcc and for what I need to compile it works just fine. That might change when and if I get Cool Threads servers in the shop, but for now gcc is fine.
Solaris 8 and 9 were OK, I prefer Solaris 10. The only time I had any trouble compiling anything with Solaris 10 is something that was built using either the entire GNU tool chain or specifically made for Linux.
While Sun points at SystemTap as a much lesser DTrace, keep in mind that at one time there was as much rhetoric coming out of RedHat against Sun as there was Sun rhetoric against RedHat. The post where I mentioned the RedHat sales droid is true, this was an actual conversation I had with a RedHat sales rep. And this is not the only example I can quote, it seemed that every meeting I went to where there was a RedHat presence, they always had something bad to say about Sun. I never heard that kind of talk from any Sun employee in face to face meetings.
If I was going to pay attention to a Linux distro, it would be SuSe. All Novell has to do is get out of their IBM OS/2 style marketing (in other words don't market it at all) and actually push their product. Their management software is clearly superior to anything RedHat has to offer, and while I haven't priced it out, I wouldn't be adverse to speaking to Novell formally.
Solaris for me works, but so did AIX and HP-UX. You just have to get familiar with each OS and their quirks, some have more quirks than others.
Member since:2006-05-13
You know you are too quick to label people anti-Sun.
You seem to think I am anti-Sun, but when Sun canceled x86 Solaris I did find McNealy's internal email address and did take the time to write to and talk to as many people at Sun as possible to get that thing back.
You simply focus way too much on what "side" people are on, and that is detrimental to Sun to have complete zealots being advocates.
And about the python experiment, hell, if someone just runs "time" on something without profiling the code, and it runs faster, say on Linux, maybe some people don't have the know how to figure out why its slower.
If every Sun-lovers riposte is to savage someone, say "where is the details" and deny any form of useful help and make a regular Joe feel like a moron for relying on empirical data, then people just go and use something else.
Arun simply did not believe I had found and filed bugs involving ZFS under load. He refused to lend any credence to the issue unless I gave him bugIDs. I know what they are, and could even provide core files. In fact, at one point, the core files I was getting were themselves corrupted, that's the depth of the issue I was encountering, yet, somehow Arun felt it necessary to simple try and assault those who would find a problem with a Sun product.
Member since:2006-05-13
I disagree with SuSE, CentOS/RHEL is a far better starting point for enterprise Linux.
I've noticed a lot of government agencies using SuSE and when I have to wedge a produce into SuSE, the build environment is really substandard, I have things building on Gentoo, RHEL, CENTOS, Debian/Ubuntu style systems and SuSE will choke. So with SuSE I usually have to make a different make file.
YAST2 is a better all-in-one configuration-station, but it is fraught with issues which makes it easier to deal with if it were broken up into separate utilities that attempted to do less.
Member since:2006-05-13
I have always used gcc and for what I need to compile it works just fine. That might change when and if I get Cool Threads servers in the shop, but for now gcc is fine.
A guy who charges $350/hr cant even be bothered to use studio. I wonder, does using different compiler than the one that compiler your OS, along with linking to the OS c-library instead of libgcc, well, wouldn't that potentially affect the OS's "trusted" or ability to meet certain criteria the government has set forth, I mean, you can just change tool-chians and put random binaries compiled with random un-vetted compilers and still be secure? Sounds haphazard.
Member since:
2008-06-05
I didn't see any trolling. He did a - harsh, but that's his right - meta-review and everyone jumped on him for his anti-Sun attitude rather than addressing points he raised.
I was curious so I googled for "shaman osnews" and found this comment in a thread: http://osnews.com/thread?231568
Have to say it rings true from what I've seen. Myself and many others commenting I'm sure are not Sun haters. Rather we use Sun products, miss the days when Sun stuff was shiny and high-performance and cutting edge and exciting. Sun seems to be falteringly moving in that direction again which is great, but this hostility toward non-Solaris users or anyone that finds something they personally don't like in Solaris is getting tiresome.
From the article linked which started that thread - http://www.regdeveloper.co.uk/2007/04/13/readers_feedback_linux_vs_... - here's a true comment. "Well, one reason that people might choose to miss out on OpenSolaris is because we're (in general) a conservative lot -- once bitten, twice shy -- and a lot of people have had bad experiences with Solaris (and, dare we say it, also with Windows and Linux) in the past. No matter how much software and UI improves, it takes ages for the community to accept this. A reputation that took years to build can be lost with one bad release -- but won't be quickly reinstated with one good one. So there will always be people who resist change -- and why not, if what they have now works for them."
Sun would do well to print that comment up as a poster and put it in the hallways. Sun has been hostile to developers since the first days of Solaris, when they threw BSD under the bus and made everything SysV. POSIX_ME_HARDER indeed. That sure worked out well, look how Sun has retained compatibility with the wide variety of Unixes out there while no one uses Linux or FreeBSD because of their oddness.
Sun's compilers are top-notch and even have a few advantages over GCC in some areas. But they kept it out of the OS for years and years. Now it's free anyway, what was the point? But Sun people complain that this code or that code only builds in GCC, or somebody linked their package against libgcc. What did they expect? GCC may be a least common denominator but if it's available everywhere and Sun won't even let people look at their kit, no one's going to bother. I'm sure people are trying to retarget Sun's compiler now but it's not going to be a high priority and it will take a long time.
Solaris 8 was actually probably the easiest Solaris to build third party stuff against. Because it had nothing. Solaris 9 and 10 include pointless old versions of things to taunt you and screw up the dynamic linker.
X server with no features. No ssh for the longest time and then when it arrived it had no features. Twitchy curses with no features. Inability to even decide what the backspace key does. This may be changing but SunOS has been a desert since > 4.1.4 and perhaps (hopefully) < 5.11. Sun and Sun advocates need to realize this and come to the game with some humility.
Instead we have Sun developers ripping chunks out of third-party software and replacing them with their own code. Their right, but when their own code breaks I have it on good authority that the internal discussion doesn't focus on "should we have replaced this" and "what did we break" but complaining about the original developer, whose code didn't break while Sun's did.
Instead we have Sun mocking Linux for years, Bill Joy taking time out from anti-nanobots rants to mock open source and free software, Sun spreading FUD about the GPL any chance they got, Sun preventing people from even doing their own closed source Java builds for free and thereby providing Sun products to more platforms. Now Sun tries to pretend they have always been #1 open source leader? Please.
Instead we have Sun acting like they are the servant of the Secret Fire, wielder of the flame of POSIX. Any good idea from anywhere else, can't have it, either because it wasn't predicted or because Sun didn't invent it. Well that is a recipe for stagnation and the world moves on, finding new standards. GNU for example: you can build that toolchain anywhere and suddenly the underlying OS is basically irrelevant. I deplore a monoculture and homogenity but much of that is Sun's fault. And tell me this, if compatibility is so important how come Solaris /bin/sh is so broken, have to look in /usr/xpg to conform; how come SMF won't read inetd.conf for legacy, even though it reads /etc/rc* just fine?
Instead we have Sun people giving talks about DTrace at conferences and wasting time making fun of Linux for exploring another competing strategy. If Sun's is so good, just put it out there and let people find it. People found Java despite Sun's best efforts. Sniping from the side implies insecurity.
Anyway. I have Solaris systems. I've liked aspects of Solaris for a long time and always respected the kernel. I hope Solaris 11 when it comes out will meet my needs and those of my users. Sun should focus on retaining and growing its current user base, and on resolving issues from beta users rather than trying to refute them.