Slashdot Mirror
The Future of OpenSolaris Revealed
ywlke writes "A few hours ago, an internal Oracle memo was leaked to the osol-discuss mailing list at opensolaris.org. It details Oracle's plans for Solaris and OpenSolaris; namely that OpenSolaris, the distribution, is dead. Solaris Express has come back from the grave, and source code will still be CDDL, but won't be released to the public until some time after it is incorporated into a binary release. What happens to the community now is anybody's guess."
The full text of the memo is available on the mailing list, as well as apparent confirmation from an Oracle employee. That said, no official announcement has yet been made.
Never mind.
RIP America
July 4, 1776 - September 11, 2001
So much for that ultimatum:
http://developers.slashdot.org/story/10/07/14/1448209/OpenSolaris-Governing-Board-Closing-Shop
Oh Oracle, what do you have up your sleeve next? Maybe you'll want to change the spelling of "MySQL" to "MY! SQL"?
There certainly wasn't a "community" for it. The vast majority were Sun employees doing their job. Linux trounced Solaris because everyone could play, Sun took way too long to realize this. No one is surprised Oracle is doing this, they make money from being an expensive closed shop. It'll be interesting to see what happens with InnoDB and MySQL in the coming months/years. Oracle are suing Google over JAVA, making people in that environment rather nervous too.
ZFS seemed pretty interesting. Btrfs might catch up eventually, but for now it's a loss.
That said, I don't think ZFS was going anywhere anyways. It's incompatible license meant it wasn't ever going to get going in Linux, and Linux has far too much momentum for OpenSolaris to have dethroned it as the open source world's golden boy.
In short the good features of OpenSolaris aren't going to have to be reimplemented, but since we were going to have to do that anyways then it's less disheartening.
"People who think they know everything are very annoying to those of us who do."-Mark Twain
From the memo:
That's wrong in so many ways it makes my brain hurt.
Maybe there's a secret footnote showing that 40% of the enterprise customers which are not currently running Solaris are willing to try it -- that'd work out nicely to 60% growth.
But somehow I doubt it.
I was just reading on wikipedia last night that OpenOffice.org is a "limited" version of the office suite, and that most Linux installs (like Ubuntu) actually come with Go O-O instead because it offers full *.docx functionality that OpenOffice.org does not. Is that true?
If so I've been recommending the wrong office suite to friends, coworkers.
"I disapprove of what you say, but I will defend to the death your right to say it." - historian Evelyn Beatrice Hall
correct. zfs was the only thing I cared about (for home use) on solaris.
its 'ok' on freebsd but not all that fast (in my experience, compared to linux md-raid, which I do realize is not at all the same exact thing).
but solaris was THE de-facto reference implementation of zfs.
kind of sorry to lose that. the rest: meh, no great loss to non-enterprise computing. and enterprise computing will still be buying solaris when they need this level of features and support (mostly the support side).
--
"It is now safe to switch off your computer."
It'll be interesting to see what happens with InnoDB and MySQL in the coming months/years.
IMHO? MySQL is in for a long, slow, drawn out slide into obscurity. Oracle isn't going to do much with it at the risk of making a free competitor to their flagship product even better.
It has been forked already, and I'm sure more will sprout up. MAYBE one of those will take off, but my guess is that without the brand recognition of MySQL to go behind them, PostgreSQL will slurp up a lot of those users.
That said, for better or for worse (worse IMHO, but that's just my opinion) "no-SQL" databases like CouchDB and MongoDB seem to be gaining a lot of traction. They won't likely take over completely as there are some things that just work better in a traditional relational database, but my guess is that a lot of smaller projects that once would have used MySQL will be looking at those instead.
"People who think they know everything are very annoying to those of us who do."-Mark Twain
OpenSolaris distributions were a joke. They would have been fine back in the 90s when it was acceptable for a free UNIX to feel unpolished, incomplete and buggy because even the commercial ones were that way.
Now with other free (as in cost) clones feeling polished and professional, and OSX being user friendly and pretty, theres absolutely no execuse for a company to allow someething like OpenSolaris to exist.
All OpenSolaris ever did was make me feel like Solaris was going backwards rather than forwards, I'm pretty sure I never had an install that 'worked' properly, there was ALWAYS something wrong. Same hardware runs Linux and FreeBSD fine, so its not the hardwares fault. My fault ... maybe, but considering I used to admin solaris boxes a few years back its not like I was completely clueless.
If Solaris Express feels like it used to feel in relation to everything it had around it, then it'll be a great improvement.
The only reasons I would use Solaris at this point are:
I want to use high end Sun hardware, meh, probably unlikely at this point.
I want a UNIX that doesn't feel like it was thrown together by a bunch of people on the Internet, a coherent experience.
I would run Solaris for the same reason I run Mac OSX, I want a professional feeling polished OS. I want to get things done, not play UNIX admin to accomplish what should be trivial tasks. The only time I should see a commandline is when I need to do something completely out of the ordinary.
Sadly, it seems that Linux's popularity killed Solaris, not because one was better or worse than the other, but because Solaris tried to act like it was Linux and just failed completely because Linux's real advantage is the surprising number of people that treat it like a god, they are a useful resource as we all know. No one will probably ever feel that way about Solaris so its just never going to get the support Linux gets from people without it having SOMETHING Linux doesn't have.
Persistent Volume manager for Kubernetes - https://github.com/dwimsey/openshift-pvmanager
There are some excellent technologies in OpenSolaris, and it appears The Illumos Project is going to be the place to find them.
I'm not sure this is a bad thing. Oracle's played its hand, and as opposed to Sun's years of "oh, gosh, we don't know if we want to be open or not - how about almost-open?" Oracle said, "screw you guys, we're going to make money off this thing." I frankly don't care about them not releasing an OpenSolaris binary build - Linus doesn't post binary builds - but keeping the source changes secret until after the commercial release just doesn't deal with the realities of Internet Time.
But, because of Oracle's decisiveness, the ON stack, the libc, etc. are all being done right now. I've tried once or twice to contribute to Nexenta and got stuck in the complexity of rebuilding a kernel, despite having done so in linux forever (to be fair the Nexenta guys were awesomely responsive so I didn't really have to do the build myself). This should be fixed.
It might give the OpenSolaris^W Illumos community a chance to succeed, being actually open.
My God, it's Full of Source!
OUTSIDE_IP=$(dig +short my.ip @outsideip.net)
That said, I don't think ZFS was going anywhere anyways. It's incompatible license meant it wasn't ever going to get going in Linux, and Linux has far too much momentum for OpenSolaris to have dethroned it as the open source world's golden boy.
Actually the ZFS storage layer was recently ported to Linux. You can use it with Lustre today, perhaps some databases. The POSIX layer is being worked on.
Due to the licensing conflict, distribution is an open problem. Probably end-users will need to install this themselves.
My God, it's Full of Source!
OUTSIDE_IP=$(dig +short my.ip @outsideip.net)
"commercial operating system" - you mean proprietary. There's a lot of "commerce" in the Linux/Free Software/Open Source world, you may have noticed it.
Augh! Posts like this make my BRAIN HURT!!
MySQL is in for a long, slow, drawn out slide into obscurity.
I agree, but not for the reasons you state. Brand recognition? Seriously? You think 30 seconds with a google search isn't going to turn up the forks?
It has been forked already, and I'm sure more will sprout up.
Except that *all* these forks have a consistent problem: there is no commercial license available. The owners of MySQL could dual-license their works, and people are free to fork the MySQL GPL edition, but they can't then turn around and offer commercial licenses to those who need them. The GPL is a bit "too free" (or too restrictive, depending on your definition of free) to be palatable.
In a strange sort of way, if Oracle doesn't develop MySQL enough, more projects will start with PostgreSQL and will never even consider Oracle. The fact that MySQL sucks as bad as it does works for Oracle, and if they actually kill it, they risk losing revenue!
That said, for better or for worse (worse IMHO, but that's just my opinion) "no-SQL" databases like CouchDB and MongoDB seem to be gaining a lot of traction.
No-SQL is not a database, it's a file store. Calling them a database is an insult to databasses the world over. Yes, there are times when a "no-sql" solution is better than SQL, and the vector is pretty much that point where you realize that storing files in databases makes sense like hauling bales of hay in sports cars does.
I have no problem with your religion until you decide it's reason to deprive others of the truth.
Btrfs might catch up eventually
Btrfs is a product of Oracle. Oracle now owns ZFS outright and controls the fate of Btrfs in terms of developer resources. One guess as to whether Oracle will remain motivated to complete Btrfs.
Oracle controls the fate of the best open source advanced file systems.
Lurking at the bottom of the gravity well, getting old
I don't think "too different from MySQL" is necessarily a minus. There's very little worthwhile about MySQL, all it had was good marketing and a earlier move to being cross-platform (which is very very important, but as a difference it's gone).
For every problem, there is at least one solution that is simple, neat, and wrong.
Ahh, the databass, such a noble fish.
For every problem, there is at least one solution that is simple, neat, and wrong.
Watch Derby. Small footprint, backed by IBM, some very nice features indeed (efficient backups and table compression can be called while running) and, although it is actually 100% java you do not need java to run it. It is a very nice way to run small, simple databases (like MySQL 3.2x was designed for), but with features like efficient complex joins and easy window selects. Oh yes, and there's a commercial version (Cloudscape). Oracle faffing with MySQL is a gift to IBM.
From scarped cliff or quarried stone she cries "A thousand types are gone, I care for nothing, no not one."
Knowing Oracle it was obvious from the day the acquisition was announced that:
1) Oracle will cripple, keep on life support or close-source all open source projects. Larry believes anything users want to use is worth making them pay for. Any open source projects that survive will be strategically useful (like letting a 'free' MySQL contaminate Microsoft's low-midrange database business revenue)
2) Java is what Oracle really wanted in Sun acquisition (see announcement today of lawsuit against Google re Android Java use) and Solaris is useful only insofar as it is part of the value prop for selling Sun, now Oracle, hardware. Solaris will only be pushed by Oracle on non-Oracle hardware if they can make a good license business out of it. Expect that all use of Java in open source implementations will dry up and any commercial implementations will be expected to start pushing license dollars back to Oracle (Which is why somebody at IBM should have been shot for blowing the Sun acquisition over the few measly millions they were fighting over before Oracle pulled the rug out form under IBM -it could have been Oracle kneeling in front of IBM instead of IBM watching the underlying architecture of Websphere and everything else Java based owned by their biggest competitor)
3) Open Solaris was a way to enable a user community (not really a dev community like Linux has) but since it can't be licensed (for money) and there's no really support/services business and it certainly doesn't help sell any Sun/Oracle hardware (which generally always runs the commercial Solaris) it has no place in an Oracle world.
I'm amazed that anybody is surprised.
its 'ok' on freebsd but not all that fast
That's an understatement. Some of the performance metrics on FreeBSD 8.1 ZFS are so poor that they're not even comparable to OSol. A 10th the performance, maybe?
Nevermind the FreeBSD implementation is shoddy, at best in terms of stability and hardware utilization in other areas: high CPU, high memory use, a couple versions behind 'official' ZFS, inexplicable instability (particularly when the filesystem is nearing capacity, but I had my test fbsd zfs system reboot itself - twice - during bonnie++ tests), and a handful of other matters.
And no, don't tell me "it'll be fixed in the next version via higher pool version support". Fix what you did before implementing something new.
Each new major version of FreeBSD since 6 seems to have taken a couple steps back where there shouldn't have been change until it worked (USB, I'm looking at you). FreeBSD is awesome for network devices and code projects, but it's kinda a wretched nightmare as a general purpose or storage OS.
ZFS in OpenSolaris is a huge loss. I just hope it's continued onward - albeit a little bit behind "official" solaris - in Nexenta and the other derivative projects. Is that even possible, legally speaking?
~/ssh slashdot.org ssh: connect to host slashdot.org port 22: too many beers
Recent Ubuntu's ship with an OpenOffice from go-oo - why do you think otherwise (perhaps there's a source I've overlooked)? If you dig into the Ubuntu Lucid source for OpenOffice.org you will see it claims the upstream is go-oo and contains many patches (SVG support, write support for DOCX etc) from go-oo. A quick web search shows the Ubuntu OpenOffice maintainer says Ubuntu's OOo is based off go-oo. This has probably been the case since at least Ubuntu 8.10 (possibly earlier).
Oracle controls the fate of the best open source advanced file systems.
If they control the fate, you can't really call them open can you?
Your hair look like poop, Bob! - Wanker.
Yes. Everyone else thinks the correct time was several years ago.
I am TheRaven on Soylent News
Alas, poor Solaris!
I knew it, McNealy, an o/s of infinite capability, of most excellent fancy.
It hath bore my applications on its back a thousand times, and now how abhorr'd in my imagination it is!
My gorge rises at it.
Postgre is ok
I beg to differ. Postgres is not just "ok" - have you looked at its features and their completeness; standards compliance; scalability (clustering); RBAC; programming flexibility; reliability? If you are a developer - how about size and quality of code, optimizer, query execution flow? Postgres probably has one of the best maintained codebase for a complex piece of software you'll ever see.
In none of the categories above can you even start placing MySQL in the same ballpark as Postgres. It's not even the same league, it's not even the same sport. So, the other part of your sentence is right in a way - it's completely different in these and many other regards from MySQL.
ZFS is already available on Linux as a user-space filesystem (http://zfs-fuse.net/) - not fast but quite functional.
FreeBSD 8.1 has the best ZFS implementation outside the Solaris kernel at present - not as recent as the Solaris ZFS but it appears to work pretty well. People who want a really point and click install for evaluation or use at home should try PC-BSD 8.1, which is a repackaged version of FreeBSD with GUI installer and simpler package installation, and is still FreeBSD under the covers - see http://www.pcbsd.org/
However, no matter how great ZFS is, you still need full backups of your ZFS storage, because there are occasions where it refuses to open the storage (zpool) and it has no fsck, by design. I like the design and features, particularly the per-block checksums, media scrubbing and solving the RAID5 write hole (http://en.wikipedia.org/wiki/Standard_RAID_levels#RAID_5_disk_failure_rate), and low cost snapshots - but the 'no data loss by design' ignores the inevitable bugs that do occasionally cause data loss.
I really hope that Debian/kFreeBSD pulls through. I love ZFS. I honestly couldn't imagine going to another file system, ever, for my server needs.
I run Xen, all of my Xen disks are just zvols. I accidentally screwed one up, just rolled it back to the last version. Because of the deduplication, I only 'used' the data that had changed.
Since it's a server, I guess I'll be one of the last to turn the lights out when something finally comes along to replace it. Xen was cake to get running (compared to Linux). It runs a Debian machine and an XP machine. Seemed adequately fast.
Except that *all* these forks have a consistent problem: there is no commercial license available. [...] The GPL is a bit "too free" (or too restrictive, depending on your definition of free) to be palatable.
Is that you, Monty?
The fact that MySQL sucks as bad as it does...
Ok, I guess not :)
I'm running ZFS with Solaris 10 on a SAN, and while I really like ZFS, I'm anxiously awaiting btrfs and will migrate to Linux the moment btrfs hits stable in RHEL 6. ZFS is good, but that doesn't mean that other file systems like btrfs don't have the potential to be better and cheaper.
Btrfs is a product of Oracle. Oracle now owns ZFS outright and controls the fate of Btrfs in terms of developer resources. One guess as to whether Oracle will remain motivated to complete Btrfs.
If Oracle for whatever reason decides to stop investing in BTRFS, the likely outcome AFAICS is not that BTRFS dies, but rather that Chris Mason and his team jump shop to Red Hat, Novell, Google, IBM or some other Linux contributor with an interest in seeing BTRFS succeed. That's one of the advantages of a collaborative project like Linux which isn't subject to the whims of any single corporation in complete control.
To the extent that there might be a threat against BTRFS, depends on how the ZFS-WAFL lawsuit plays out. I wouldn't be particularly surprised if Oracle settles with Netapp, covering only official Solaris releases, leaving other ZFS versions (Illumos, Nexenta, FreeBSD, etc.) out in the cold, and perhaps BTRFS as well, depending on to which extent the WAFL patents apply to BTRFS.
The reason MyISAM (not the whole MySQL) is "fast" is because there is no proper abstraction level between what executes the query and the actual file writes. This is true for simple inserts, simple selects, simple updates. This is also why when MySQL crashes, quite frequently MyISAM tables become corrupt - you can try to repair them, but hopefully you were replicating.
Besides, you use MyISAM for "speed" and you lose basic functionality like transactions, MVCC, ACID compliance (hmm, did you even have it in the first place?), row/page locks, etc. You can perform direct file writes even faster than that, I guess it counts for something, but that doesn't do you any good either.
On the other hand, what kind of "real-world" benchmarks did you do? No such "real world" I know of consists of simple inserts and selects. How about cases for:
- optimized subqueries
- using index merges
- reusing indexes in same query
- partial indexes
- indexes on expressions
- transactions with savepoints
- etc., etc.
MySQL doesn't do any of the above. Welcome to the "real world."