The Eclipse of Sun

If you've been reading this blog for a while you'll know that
for the past decade I've had a love-hate relation ship with Sun.

Sun has done a lot of good things in their open source journey such as MySQL, openOffice , Virtualbox and opensourcing Java. But on the other hand they had troubles with creating really organic opensource.

I've always wondered why a company that bought StarDivision because it was cheaper than buying the MS Office license keeps wasting money on building it's own operating systems, albeit in a pretty open model, but with zero to no relevant community.

Yes I've ranted on their marketing boys and girls when they polluted forums I liked. But no I don't want this to happen. (Slicing up Sun)

However I`m in fear it actually might happen, the funny thing is that I already mentioned the same idea in offline conversations during the last couple of months.

My Advice to Sun. : Stop putting money in things you can co-create with other organisations, if there's only 2 features in your product that count for some people, contribute those 2 features to something bigger.
(That's D-Trace and ZFS and the Linux Kernel I`m talking about in case you didn't realise) Why keep throwing money down the drain ?

Also let Google, Percona and others contribute to MySQL, don't keep it as closed as it is today. As after al it's more Free software than it is Open Source, but it has a great community which I love to be part of .. and I hope for a long time to come..

I`m going to try to keep my mouth shut on this for a while.. as I think Jonathan is listening anyhow. If he was, he would have known that at some point in time when you step out of the shower and there is too much hair dropping from your skull every morning, it's time to cut off the ponytail.. , not opensource it. I did that years ago.. and it's growing much thicker again now.

My hair that is ..

Comments

Dieter Van de Walle's picture

#1 Dieter Van de Walle : Zones

What about the Solaris zones and LDOM's?

I've been looking at OpenSolaris for a few weeks now and it seems véry cool. Not easy if you're coming from Debian/Ubuntu though ...
Not much documentation to find.
(Apart from the gazillion pages official Sun manuals...)

I think I understand why Sun is keeping tight control over Solaris though. Features like D-Trace, ZFS, zones, LDOM's seem to me as very deeply integrated into the system. It doesn't seem to me that these features are so easy to just add to the existing linux kernel?

Same feeling on the love-hate relation though ..


Kris Buytaert's picture

#2 Kris Buytaert : Tell me

Why would I want Zones when I already have Linux Vserver or openVZ to choose from ...

Let me know.. I might learn from it ..


Matthew Montgomery's picture

#3 Matthew Montgomery : MySQL Contributions

"Also let Google, Percona and others contribute to MySQL,"

I think you're missing a key piece of information here, most of the Google and Percona patches apply to the InnoDB storage engine. Heikii/Oracle has refused to include performance improvement patches in the current branch of the InnoDB engine. Sun/MySQL is prohibited by contract with Oracle from using anything other than the mainline branch of the engine. Really, a good bit malice over acceptance of these patches is wrongly directed. Other patches are not general enough to apply without causing problems in other parts of the server. Google's semi-synchronous replication patches are a good example of this. It only applied to InnoDB, it required significant recoding so that we could apply it to all engines. It has been accepted and will be part of the next GA.

http://dev.mysql.com/doc/refman/6.0/en/replication-semisync.html
http://forge.mysql.com/wiki/ReplicationFeatures/SemiSyncReplication


Kris Buytaert's picture

#4 Kris Buytaert : True and hmm.. well.. recent :)

Matthew, you are right about the InnoDB innovations.
I never looked at it that way .

However fact is that even before MysQL was part of Sun, or before Oracle bought InnodDB getting contributions accepted into MySQL wasn't really common. I think I even mentioned once that with this lack of massive outside contributions on both MySQL, VirtualBox and Lustre they were all perfect matches for the still way to closed company that Sun is while claiming to be an open company.

So I guess the Percona and Google examples weren't really that good :)


Matthew Montgomery's picture

#5 Matthew Montgomery : Improvements in community participation.

It has been recognized that the acceptance of community patches is still too slow. Internal parties have complained that their own patches have not accepted quickly if at all. Even prior to becoming part of Sun, MySQL had been taking some steps to improve this and make community participation in the development more of reality. This has continued following the acquisition. Some examples of these are:

1) Create a plugin interface for community extensions
2) publish worklogs publicly on forge.mysql.com
2) Change to a more free and open revision control platform. (launchpad.net/bazaar)
4) Adopt the Sun Contributor License Agreement, which is more permissive than the original MySQL Contributor Agreement.

You cannot say that MySQL has not made progress in this area.

MySQL has historically had an open source distribution model but not an open source development model. Sun has shared this dilemma. Sun produces more open source software than any other company on the planet. So, you cannot say they merely claim to be "an open source company". Sun and MySQL desperately want to further open development to outside parties. They also have a responsibility to perform releases not destabilize the products they provide, while showing profits.

I must ask what other open source companies in your mind accept "massive" outside contributions? Canonical? Yes, so we adopt their revision control system in order to become more like them. However, I would argue there are few others who take many, many patches from all but a small number of contributors.