(will there be) a future for OpenSolaris?

Saw this linked to from /. . Its pretty clear that Oracle is taking a deep, long, hard look at all projects within Sun, figuring out what to keep, and what to abandon. Things which have no hope of revenue generation, or driving business in general are not likely long for this world.
This brings us to OpenSolaris. This is the “open source” version of Solaris. I put it in quotes, as it may technically be an open source license in some manner of speaking, but it is fundamentally incompatible with GPL, with Artistic, with … you name it. The code base cannot be contributed to from a GPL or other code base. There can’t be cross pollination of efforts. Which tends to cut down significant contributions from serious companies that view adding features and functionality to GPL projects as a way forward with their product offerings.
So I read this with some trepidation.
In short, Oracle is asking OpenSolaris folks to help justify its existence as something funded.

So here’s a challenge for everyone who wants to make our product even
more Open. Show us a plan for how that will ultimately generate revenue
for Oracle? The “just do it and see what happens” idea won’t fly —
management will need to have a real plan. Ideally backed up with some
research (marketing numbers, etc.) or other evidence. Remember, this is
a real business with millions of real dollars invested in it — its not
some high school economics research project.

I read this as several things. First, a deep and fundamental misunderstanding of how open source works. Second, significant concern from those who want to retain the “open source” project going forward, that they see that they need to be productive to Oracle’s bottom line.
Oracle is a business, and it wants to make money. Pure and simple. So how can a software company make money with an open source project? I mean its … like … impossible. You give away the program, code, and other bits for free … and no one will pay you for support.
Right Redhat?
As I said, a deep and thorough misunderstanding of open source.

If giving stuff away free does lead to *revenue* (and this was *part* of
what an Open Source Solaris product was about — enabling sales of Sun
hardware), then great. (The other part was an attempt to contribute to
breaking the Windows monopoly on the desktop. I think Linux has done
more here though.)

I’d put Solaris/OpenSolaris out as its own business unit. No hardware ties at all with any Oracle business unit. You have OpenSolaris (akin to Fedora/Centos in some regards) and Solaris. This business unit will either live, or die, based entirely upon revenue it brings in. Its business model should be effectively identical to Redhat’s. Redhat is quite profitable mind you, on software anyone can pull down and install on any number of machines. It has redistribution agreements with many hardware vendors.
So should Solaris Inc.
This model obviously works.
Do the same thing with MySQL. Commercial/Open product.
Oh, and by the way, change the licenses to GPL so people can actually contribute to it. Without that, Solaris Inc. is dead on delivery.
Without something like this, OpenSolaris, and Solaris face a very bleak future. OpenSolaris in part drove people to consider using Solaris commercially. Without OpenSolaris, commercial Solaris is pretty much toast. It was heading that way before the OpenSolaris project began.
The issue is, fundamentally, egos get and had got in the way of doing the right thing, coupled with some poor business strategic and tactical decisions.
Ask most Sun folk and they’ll tell you that Solaris is infinitely superior to Linux and everything else. Then when asked why this didn’t translate into commercial success/thrashing of the competitors/why Sun shipped more Linux than Solaris … most will shrug or offer WAGs.
I can’t tell you how many times I’ve seen this pattern … in companies I worked for (SGI) and in companies we’ve worked with (Sun).
If you are not winning or gaining market share, you are losing ground. And if you are losing ground, deluding yourself that you are better isn’t going to help you figure out why you are losing. Because you will have to climb out that much deeper of a hole after you are done, the longer you delude yourself.
This is where OpenSolaris now finds itself. The Oracle folks said “Hey, you are deluding yourself, this is not a commercial success, a driver of new business and new revenue. Show us how we are going to make money with this. Or kill it and write off the investment.”
What sad is that the path to turning it into a potentially profitable endeavor requires a focus upon doing something the Redhat does, and fixing the license.
Redhat doesn’t develop all its own product … this is done globally, and they aggregate work done by many others. They package it, they support what they package. They do some development on some critical bits, but as it is GPL, everyone else can contribute.
Notice that. Because it is GPL, everyone else can contribute.
What license is OpenSolaris using?
The fix is fairly easy to do, but likely terrifying for OpenSolaris/Solaris business folk. It requires a leap of faith. A huge one.
One that Redhat has made and continues to make. And it is doing quite well at that.
SuSE is doing reasonably well as well with this model.
Hmmm. … I wonder if they are onto something.
I guess I shouldn’t keep pointing out the obvious here. I’ve not been hired to do this, so my advice is somewhat armchair quarterbacking. But if i were hired to do this, i’d write my business plan and elevator pitch just like Solaris was a real open source startup. I’d start out with the license change, the community growth, and focus on the only thing that matters for Solaris Inc. The revenue that Solaris Inc will add to Oracle Inc going forward. Not in additional hardware sales (that was a foolish model). But in a Redhat like scenario.

4 thoughts on “(will there be) a future for OpenSolaris?”

  1. No one should be surprised about Oracle’s recent decisions and comments. They telegraphed it pretty well. I just think that people are a bit shocked at the speed at which Oracle is cutting their throat, er, losses.
    In addition to OpenSolaris, how about the recent Lustre roadmap. Looks like Lustre will be an Oracle only product starting with Lustre 2 where you have to buy their software stack (likely to include Solaris) and their hardware if you want support (I wouldn’t be surprised if they closed down the mailing lists or at the very least took their developers off the mailing list by stating they are “too busy” to provide free help on the mailing lists). There is a reseller agreement planned but I’m wondering how many companies what to be stuck with an Oracle stack that is likely to include Solaris at some level?
    I’m glad MySQL is forked because I think Oracle is going to kill it or incorporate it into Oracle at some point.
    Oracle’s overall plan is pretty simple: go closed-source, push Sparc, push closed-source software, push expensive licenses. It’s been the Oracle model for some time and and Oracle is ready to use it as the proverbial hammer when it sees everything as the proverbial nail.
    I can’t see them succeeding in HPC at all. Sparc in HPC? Tried that – the market didn’t care for it. Solaris in HPC? Tried that – the market didn’t care for it. About the only things they have going for them are Lustre and perhaps ZFS. But Btrfs is better than ZFS (and it’s supported and open-source!) and the Ceph client is already in the kernel (Lustre has never been able to make it into the mainstream kernel nor in any distros).
    Oracle in HPC – forget about it.

  2. @Jeff
    I’ve been holding off writing about this in a holistic sense, or in a detailed sense fo a little while, but I do believe you are right.
    I read the Lustre 2 roadmap as a closing of Lustre. There are options for getting support going forward, but the code base will effectively cease being OSS from what I have read.
    Your other points are spot on, but allow me to momentarily assume the thought processes of an Oracle GM on this. What fraction of Oracle revenue will HPC and the product mix, generate? What is the cost or fractional cost of this revenue? Does it make sense to continue to invest in it, or shut it down?
    This is the same issue with much of their hardware offerings, though I’ve heard some interesting analyses why they are retaining Sparc … more to do with competition from IBM than anything else.
    But, at the end of the day, they have to do this cost-benefit analysis. If the benefits don’t exceed the costs, which would be the primary justification for continuing the efforts, you have to look at the secondary and indirect benefits … does working on this and spending this money provide indirect revenue to the corporation, or goodwill, or …. If the value of this indirect revenue or other intangibles isn’t there, then the decision is obvious, though often hard to make.
    Sun could never have made such decisions. Oracle is smart enough to do this.
    HPC @ Oracle doesn’t appear to be long for this world … not just in a general sense, but in a particular project by project sense. Lustre is a decidedly HPC product. You wouldn’t want to run a database atop this.
    If Lustre is to survive, I think it needs to fork. The sense I have from this roadmap (I passed it on to a few customers who have decided to go Lustre) is that Oracle wants this fork to occur. It would enable them an easy path to say “hey, we don’t need more than 1 or 2 engineers working on this full time”, and effectively close down the rest of the work. The problem is, as a business, Lustre had failed at CFS. It isn’t viable, its a small market given the focus of Lustre. I don’t think a commercial entity would take it up on its own.
    It is sad, but it is expected. Oracle is a business, in business to make a profit for its shareholders. I don’t and won’t begrudge them the decisions they need to make to do this.

  3. @Joe,
    Totally agree. However, I think Oracle can make money from their open-source projects. As you pointed out Redhat has been able to figure it out. 🙂 But I support Oracle’s decision to make $$ decisions around their portfolio. They need to be make $$ to be successful, to stay afloat (not that they are in any danger), and to develop new products/technology. I think they can do this with open-source software in the mix.
    I also think they have (had?) a shot at doing well in HPC. The problem they face is turning it into a profitable business (not easy but doable).
    Since we’re both into HPC we can make the argument that HPC is a useful segment for companies because we encounter problems well before other aspect of the market. Plus, if you believe Intel and IDC, a big chunk of CPU sales are around HPC. So it can be very useful to companies to keep HPC around if the other parts of the business are large enough to sustain losses. Forcing individual technologies are products to have make money consistently is perhaps not the best way to run a company but I don’t think Oracle is doing this. 🙂
    Fun times for Oracle but a bit scary as well (need to make money, develop roadmaps, etc.).

  4. Oracle say Lustre 2.0 will continue to be GPL and you will be able to report bugs against it, but if you want paid support you have to run it on their hardware (though they talk about 3rd party reseller agreements, but no idea if that means on their hardware or not).
    In other news Peter Braam (one of the Lustre creators) now has a company called ClusterStor which will “develop a new distributed, clustered and parallel storage system that will surpass existing records and take the entire storage industry into a new epoch”. They also say they offer paid Lustre support.. 😉

Comments are closed.