OpenSolaris what needs to come next, or, the day Ian Murdock joined Sun
I’ve had a half-drafted blog entry sat around for about two months now. I started to write it after I did an initial install of a (at the time) bleeding-edge copy of Solaris Nevada.
For those that don’t know, Nevada is the code-name for the next version of the Solaris operating system, Solaris 11. Nevada is actively developed where all of the new features, improvements, bugs and bug fixes end up before they get putback to Solaris 10 in the form of updates or patches.
Most of Sun runs on Solaris 10 Sun Ray servers, but in the department I work we run and maintain our own servers: one runs Solaris 10 and three run Solaris Nevada. By choice I use Nevada for the more recent version of the Java Desktop System or GNOME.
In an office workstation role, Nevada/GNOME works surprisingly well. From time to time Solaris does panic, but the bug is always logged and fixes come in with future builds. At home, on a stand-alone desktop PC, both Solaris Nevada and Solaris 10 fall well short of the current mark, which I will define to be somewhere in between Windows Vista, Mac OS X and, of course, Ubuntu Feisty Fawn.
As a long-term Ubuntu user I can’t help but cringe when I look at various aspects of the normal Nevada installation, use and maintenance. Depend on moods in the office, brining these problems up elicits one of two responses:
- Yes, but… the Linux kernel sucks. There’s no stable driver API, it crashes a lot, it’s a hackers’ OS, etc. My response to this is: so what? Solaris is no use to anybody at all if it requires n hours a week/month in maintenance, doesn’t work with their hardware and can’t run their applications. Years ago when I tried to convert my Dad to Linux his reason not to was quite simply that it didn’t run his accounting software. And besides… my desktop crashes less frequently than the Sun Ray servers at work (not a fair comparison, I know).
- Yes, but… nobody is willing to do the work, it isn’t a priority, businesses don’t want it, etc. This is definitely a more reasonable response.
So what is Solaris Nevada lacking that Ubuntu already provides:
- Package management, package management, package management. For the enterprise the current solution might be acceptable, but desktop users need to be able to painlessly install updates. Basically what Solaris needs is apt-get/dpkg.
- Installation. The Solaris installation is archaic, although it does mostly work. The text interface scares newcomers (even intermediate/advanced ones) off. Nevada needs to pay close attention to what Microsoft have done with Vista and Ubuntu are doing with their live CD installers.
- Software. Ubuntu ships with a sensible set of default utilities, applications and configuration tools. For the most part, so does Nevada. But what Nevada (and Windows and OS X) doesn’t offer is a single interface to find more software for a given task. This is where I bring up (again) apt-get, but this time in conjunction with a great big repository of software, one very similar to those offered by Debian, Ubuntu, Red Hat, SuSE, etc.
- Community. Ubuntu has the most incredible community I’ve come across on the Internet. Thousands upon thousands of people of different levels of Linux experience together to help people use their computers. OpenSolaris is a great community that Sun has helped foster; using what they learned from this they need to set out and create a Desktop OpenSolaris community aimed at the everyday desktop user.
So how does Ian Murdock fit in?
Ian Murdock joined Sun on March 19th this year. Ian founded the Debian Project and was the first DPL (Debian Project Leader), he’s been hired by Sun to help with Operating System Platform Strategy. He’s already made it very clear that he wants to close the usability gap between Solaris and Linux, but also that he believes Linux needs to play an important role in what is to come.
Over the last week or so Project Indiana has been banded about quite a bit. I don’t think anybody really knows quite what Indiana has in store for us just yet—I’ve certainly heard nothing more than I can read on the Internet, despite the fact I have access to SWAN. It is entirely possible that Indiana has nothing at all to do with Ian Murdock, but the rumourmill has been set in motion, and an interesting little fact is that Murdock was born in the US state of Indiana…
What will Project Indiana be? A better question might be ‘What do I want Project Indiana to be?’ I hope that Indiana will be the start of a big push for Solaris to hit the desktop. I hope that Indiana can take over what the guys over at gnusolaris.org / Nexenta have begun:
A complete desktop operating system, based around the OpenSolaris kernel, a GNU userland and oodles of open source and third-party applications hosted in an easily accessible repository.
What if, given the fact that Sun endorse Ubuntu on their UltraSPARC T1 processors, Project Indiana became Ubuntu Indiana…
May 10th, 2007 at 8:22 pm
If you make Solaris more like Linux, then what happens to all the stuff that makes Solaris great? At a normal, day to day user the Solaris kernel has very few things going for it over Linux. Yes, they can fix a lot of the front end stuff, slap together some nice installer, get a package manager sorted, setup and foster a forum. But why switch? At the moment the massive advantage that Linux has is that a lot of things have finally started Just Working™. People switch to Linux because it’s free, it doesn’t have some of the retardedness of Windows. So, why would someone want to switch to Solaris. Sure, ZFS is very nice, so is dtrace, the solaris linker, the kernel, so on and so forth. But this is stuff that you care about at the sys-admin level, at the hacker level, not the user level.
You see, the problem now is that you’ve, not necessarily you personally, have gone ahead and evengelised Linux for years. So the people have come. And I think it’s worth noting that in a lot of cases they came because they hated MS, not because they wanted Linux. Now, having spent all this time getting comfortable you want them to switch? Suddenly, the big MS demon isn’t there to push them onto another OS, Solaris has to give them something to pull them in, and I don’t think it will happen.
Therefore, while it’s nice to sit back and hope, unfortunately Linux has successfully won the alternative OS brand, and to now shift it from there will take years, and maybe even decades.