Comments on: Ubuntu + Sun = Very good idea

Sun and Ubuntu could make for a very fruitful partnership.

[buraddo]

You idea assumes;

1. Every workload on the planet can be serviced by one kernel architecture. Certainly not possible given the current OS architectures. One workloads core features is anothers baggage. You see this 1 size fits all mentality debunked up and down the IT infrastructure stack.
2. Competition is bad - with all the momentum, adoption and community around linux, there is still a select few committers directing the architecture at the kernel level. I feel more comfortable in the long term is a few different approachs are being taken.

Embracing linux is better for linux, Sun and the IT industry. It supports some of Sun HW business and all of its software business. It creates competition in the OS market which increases its importance and business value, meaning Ubuntu and Redhat dont have a declining market value situation.

What I would like to see is some collaboration with OS architectures to eliminate the app vendor certification concept (and I am not necessarily talking binary compatibility). Certification by type and version of OS is holding our DataCenter back, creating massive inefficiency in resource usage and is one of the main reasons for the Hypervisor concept even existing.

[buraddo]

This idea assumes;

1. Every workload on the planet can be serviced by one kernel architecture. Certainly not possible given the current OS architectures. One workloads core features is anothers baggage. You see this one size fits all mentality debunked up and down the IT infrastructure stack.
2. Competition is bad - with all the momentum, adoption and community around linux, there is still a select few committers directing the architecture at the kernel level. I feel more comfortable in the long term is a few different approachs are being taken.

Embracing linux is better for linux, Sun and the IT industry. It supports some of Sun HW business and all of its software business. It creates competition in the OS market which increases its importance and business value, meaning Ubuntu and Redhat dont have a declining market value situation.

What I would like to see is some collaboration with OS architectures to eliminate the app vendor certification concept (and I am not necessarily talking binary compatibility). Certification by type and version of OS is holding our DataCenter back, creating massive inefficiency in resource usage and is one of the main reasons for the Hypervisor concept even existing.

[thevarguy]

Matt: We seem to be on the same page again and again. I think we'll start to get more answers during the Ubuntu Live conference this summer, which Sun apparently is sponsoring.
-jp, representing The VAR Guy (www.thevarguy.com)

[johnjmclaughlin]

Does Ubuntu have Dtrace, ZFS, FMA or Containers? Does it drive billions of dollars of sales in the datacenter? How many of the thousands of Solaris/SPARC apps will run on it?

According to IDC, Solaris accounts for about 50% of all UNIX server shipments - that sounds like momentum.

By the way Sun has Open Sourced Solaris and the OpenSolaris.org site has 100k registered members.

That said, Ubuntu on Sun's AMD and Intel servers will be interesting if they get the same level of ISV certification that SUSE and Redhat do.

I'd also like to point out that it is "LAMP" on commodity h/w that is popular. With the Apache, MySql, PHP/Perl/Python and a ton of other GNU s/w integrated into Solaris and compelling kernel features that, in their arrogance, the linux kernel team will not adopt (Dtrace, ZFS), I see a great deal of interest in Solaris and OpenSolaris

[drgrep]

"In a May 2006 report, IDC reported that 66.3% of the AMD-based Galaxy servers in Sun's portfolio ship with Linux pre-installed. By September, that number had increased to 71.5%. A Sun representative contacted for this article declined to update those numbers, and said the company does not break out the number of servers shipped for analyst reports."

http://searchenterpriseliux.techtarget.com/news/article/0,289142,sid39_gci1238478,00.html

Sun sells more Linux than Solaris on its own hardware. And as for Solaris, 70% of all x86 Solaris runs on HP hardware. What do both of these facts say about Sun's strategy with Solaris?

Solaris has been "open source" (but not self hosting) for three years now. Nobody cares. It's still clear that Sun doesn't want anyone creating their own OpenSolaris distribution as its all but impossible to do, and the existing developer communities who are actively involved in such efforts (not just "registered" on a Solaris Express download site) consist of about 30 people in total. The greatest achievement to date on this front has been Nexenta, which truly did create the best of both worlds: Ubuntu userland with Solaris kernel, and they've been all but ignored by Sun.

The Solaris kernel is clearly better than the Linux kernel in countless ways. It handles heavy loads more gracefully, it is more standards compliant, it has true enterprise capabilties, professional documentation, a file system (ZFS) which is shaping up to the best ever created, and what can you compare DTrace to? It's phenomenal.

But superior technology doesn't trump community. Just as PostgreSQL is starting to eat the bell curve out of Oracle's market share, so has Linux to Solaris. Why? They play well with others. They are designed to be truly open, completely non-restrictive, and community friendly. Even Linux's model is suffering some from its autocratic nature, causing some talented developers to defect to the FreeBSD project, which is more democratic and community friendly.

Look at Sun's most recent purchase -- MySQL. MySQL, while a good database, is not a technologically superior database, not historically anyway -- and everybody knows it. What was so great about MySQL? 1) It was easy, easy, easy to use, 2) It was truly open source and 3) most important -- mega community (LAMP) (which are a result of 1 and 2). Sun was not buying technology with the MySQL purchase, they were buying community (LAMP), and that says something -- that they are so incapable of creating community on their own -- even with superior technology and products -- that they have to resort to buying it for exorbitant costs (1Billion!? -- For a database that doesn't even have it's own transactional storage engine? That's what IBM paid for Informix!). How is it that MySQL can be worth the same price as Informix? It's obviously not the technology. It's the community, proving that this is what matters in the market. This is why MySQL is still here and Informix is in the IBM retirement home. It's why MySQL is more widely used than PostgreSQL, which is technologically a far superior database.

And so, if Solaris is to stay relevant, it has to really and truly yield itself to the community -- which means yielding control to others outside of Sun. That starts with creating a truly self-hosting open source OS in which *every* component is open source and redistributable, which to date is still hasn't happend. When this happens, then it means that a community can take the entire product and run away with it, even produce something new that competes with it. And this is the soul of open source community -- the freedom to take the project anywhere, or fork it. I think deep down Sun fears this, and this is why they still don't provide the means necessary to create a self-hosting, redistributable, open source Solaris. And paradoxically, this fear of theirs is the very thing that is killing it. It is Sun's very protection of Solaris that keeps the community from coming in and saving it.

But even if this were to happen, many have argued that it's too late. People are happy enough with Linux in 90% of their applications. Existing companies have hired Linux and BSD kernel developers and have momentum developing drivers and other features. Linux/BSD are very well-established. They have come to be standard parts of major enterprises -- Yahoo, Apple, IBM, Cisco, Google, HP. It is extremely doubtful at this point that even a successful, community-friendly OpenSolaris has any chance of displacing Linux or BSD in their present establishments. All that's really left for Solaris is just huge, high-end deployments -- and where's the community in that? That's the domain of about 5% of the market, which consists of Solaris certified admins who can afford expensive support contracts for their mainframes.

On its present course, Solaris will not die, but it will become what AIX and the z10 mainframe are today, a critical but very small part of the market, always with a persistently uncertain future.

And while you might argue that Linux doesn't have ZFS and DTrace, well FreeBSD is pretty much inheriting these crown jewels. And that's the sad thing: Sun, while not going far enough in open source to really build a community, has unfortunately gone just far enough to allow others to incorporate the best of their technology. FreeBSD 7 is the advent of massively scalable open source UNIX, and scales linearly all the way up to 8 processors, and is even more community friendly than Linux.

In the end, open source will prevail. The only question is whether Sun will be one of the parties benefiting from it. Going open source requires absolute and total commitment. Doing it half-heartedly is worse than if you never did it at all. If Sun doesn't let Solaris walk out of the company, which is ultimately the essential freedom needed to draw developers to it, then it will simply be stripped of its parts, which will in turn go to making other OSes even better, as is the case with FreeBSD and OSX today: both of which offer ZFS, with DTrace next up on the feature list.

I personally wish the Solaris kernel would replace the Linux kernel. It's a thing of beauty -- true engineering at its best. But so were products from DEC and SGI.

2 years ago:

http://blogs.sun.com/jonathan/entry/ubuntu_on_niagara_and_platinum

[hackertarget]

Ubuntu has a cool factor amoung the kids and a rock solid debian base. It is unstoppable at the moment in the desktop space and is not far away from becoming a big thing in the server space.

Sun could only benefit from such a well known and popular brand.

[angrykeyboarder]

I love Ubuntu, but Solaris is also a great OS. And you really to to take a look at OpenSolaris' Project Indiana. It's being run by Ian Murdock of DebIAN fame).

By the time they get to teir first release it will be as easy to install as Ubuntu. And being that it's Solaris it will have all the previously mentioned features that Linux does not have.

A lot changes quickly in the tech business. You never know, someday OpenSolaris may be where Linux is today.