This fact is sad but true, Solaris is pretty much a legacy OS. In many ways Solaris is technically superior--but it doesn't matter. The best technology doesn't always win, just the one that has the most momentum and largest user base.

I remember when Sun usurped SGIs role in telecom and financial and now it's happening with Linux. The real question is what's next?

Matt,

Sun already has the next big thing. Massive multi-threading chips. Bank of America is on-board and will do a big roll out of them to drive home a 'Higher Standards' message around environmental responsibility.

The best part is no one will see them coming until they are already back.

Rumors of their demise have been greatly.....

James

hi there,

The licensing of OpenSolaris is partly responsible. It's been 2 years since sun announced their decision to open up the source. I have not heard of any decent community build up for this distro.

The licensing is not good -- that's why most of the *BSD products are not making money for any vendor.

The only recourse available for SUN to save Solaris is to GPL Solaris and put thousands of sun engineers and marketing people to build a company around a GPL'ed Solaris -- borrow lots of programs and hardware detection code from Linux.

Linux's mantra was : better UNIX than UNIX. Now Solari's mantra should be : better LINUX than LINUX!! :-)

I know for a pioneering company like Sun, it will be a hard swallow but unless the environment changes dramatically, a GPL'ed Solaris is the only way Sun can Survive the Linux onslaught.

BR,
~A

How do you explain over 5M+ Solaris 10 downloads...momentum is only building

I agree with James Walter on the chips first and then I really think open sourcing Java is right there too. Maybe Solaris third.

The OpenSPARK community is great and the UltraSPARK chips are under the GPL too.
So as a small business, I can repair the chips or inmnovate them at a much lower cost instaed of Pay-to-Play politics.
Plus ist's great for the environment so cips are not disposed of too much.

The problem with writing something is that no one seems to read it. I think Solaris may well have a healthy future (just as Novell continues to pull in $1B/year on old NetWare licenses, among other things), but I don't think it's a revenue stream for Sun going forward. Not in the way it has been. Sun hardware (as has been suggested above in the comments)? It's awesome, and will continue to find a buying audience. But I just don't think Solaris is going to be a big money maker for Sun, and I think at some point Linux eats its lunch.

Well, for our take, we switched to linux for development and testing about 18 months ago. It was a pretty good experience, nothing horrible happened, things were compatible pretty much, our distro (RedHat) was stable and administration was OK. But we're making the shift back to Solaris now because it's just a bit more solid. When you really (really) push your gear it makes a difference.

Oh sorry, I read it, but it's a slightly complicated subject as in someone running your hardware can encourage them to want to use your software like SPARK with Solaris. So opening SPARK might have a good effect even if they are x86ing it with Linux too. If Sun can stay being a successful hardware company by embracing the GPL on hardware then it's hard for me to see them switching to Linux.

Guys, OpenSolaris performs ~50% better than Redhat when using MySQL. Facts are facts. Watch out. Here are the reports...

New OLTP Benchmark Results for MySQL on Solaris 10

http://www.mysql.com/news-and-events/press-release/release_2006_19.html

http://www.sun.com/x64/docs/MySQL-sysbench-benchmark.pdf

The Solaris advantage was magnified during the read-only test, where performance exceeded the Linux test case by 91 percent. Remarkably, in this experiment, the peak performance under the Solaris 10 OS was achieved with 16 CUC, while the less robust Red Hat Linux tapered off at only eight CUC. Despite running at twice the load during the peak phase, the Solaris 10-based server was performing 53 percent more transactions per second than the Linux-based server.

As long as Solaris on x86 is able to run off-the-shelf binary software that has been written for popular x86 versions of Linux (i.e., RedHat), then I'd opt for Solaris as it's a much better Unix-style OS with some nice advanced features.

Linux is so over-rated. Mac OS X on the desktop kicks it around and makes it look like an OS written by preschoolers.

Linux GUI vs Mac OS X GUI

I much prefer to use the best stuff and open source is by no means necessarily the best. For instance, Tibco EMS (an enterprise implementation of JMS messaging) is many leagues superior and more solid that open source ActiveMQ - which is the leading open source JMS.

As far as cost - for enterprise developers there's not that much difference. When you go to sign up for annual support contracts you'll find out that the open source guys are right in the same ballpark as the proprietary alternatives.

The thing the open source products have going for them is a marketing advantage. You get to kick the tires by using their software for free until you decide to adopt it officially and then sign up for support. But at that point there is no cost advantage.

I dwell on open source because for a lot of enterprise class products it's not really amounting to that much of a difference other than this marketing factor. JBoss Inc. relies on its revenues to fund its R&D because only its staff engineers work on and develop its software. There is no community out there that does development with them - just a community of users. So in the end the JBoss business model is no different than the traditional proprietary models that came before it other than the marketing advantage I just cited.

And that marketing advantage can be rather minimal. For instance I can download a version of the Tibco EMS server that is fully functional and that I can do all manner of real JMS development with. It's only limitation is that it will not run for more than 24 hours before it exits. This is hardly a limitation for a developer. All of the vendors that my company integrates via our use of Tibco EMS never have any issues with using this "crippled" server to do their development with. Then we deploy the uncrippled version of the server at our production sites as we pay for licensing and support.

There's really marginal difference in this vs. my ability to download and use any of the JBoss JEMS products for free.

In the end when I put any of this software into production for my company's enterprise operations, the cost of either proprietary vs. open source will be pretty much a wash.

The whole commercialization of open source into the business sphere is thus this gigantic chimera that the technology media have latched onto in order to spin endless articles pontificating about it.

The reality of the pocket book - when it gets to that stage - nothing has really changed.

Nor in the end has anything changed in how software is developed.

Communities of volunteers only develop the lessor stuff like, say, the JDOM XML library. And a lot of it is funded in some manner, such as the govt. funding behind the Java JAIN SIP reference implementation library.

Communities are incapable of producing an enterprise class of software of the range and level of sophistication of the likes of the JBoss JEMS stack. Only professional, paid software engineering staff can do that. It is entirely impossible for a volunteer community to develop software that is as advanced and aesthetically pleasing as Apple's GUI for Mac OS X. Only Apple's paid talent is up to snuff for pulling that off. The open source portion of Mac OS X is the underlying 25 year old Unix-like kernel stuff that has been well trod.

I'm rather amused at the things articles such as this come out and claim. Sure there is layer of the tech. industry that are pop culture driven by faddish phenomena - such as open source and Linux. But the underlying core of corporate IT is much more sanguine and concerned about the actual real-world reality vs. the hype.

Now that we've had a few years of deploying open source software in the enterprise along side our proprietary stuff, we now see that the alleged difference is entirely media hype fabrication. The real world reality is that commercial open source is actually very much the same as the proprietary stuff.

Yay, who needs an OS that can handle massive threads in a world where single thread performance continues to decrease and multi-core multi-thread becomes the norm. Just give me an OS that everyone perceives to be cool, perception is reality for those who dont bother to actually understand what they are talking about.