Crossbow - Network Virtualization Architecture Comes to Life

I like this Solaris project. It was just a practical issue for me: being able to virtualize network devices. Almost everyone had this feature, more or less. But this is perfect. You can actually do traffic shaping, dedicate cores, etc. This is what every virtual hoster would love.

But it also shows something I don’t like about Sun. It’s hype about technology, that does not even exist yet. It began with Solaris 10. We have Zones and ZFS, but just not yet. But they are cool! But we can not see them, they are in development.

We have JavaFx! Glassfish 3! Open Storage!

SAM-QFS will be open-source!

Sun xVM server will (some day) be answer to your virtualization problems!

It’s just not happening and I don’t really get why Sun has to spread the hype about technologies, that do not exists yet. They create hype, but with that comes disappointment, that these cool technologies do not yet exists outside of Sun laboratories. People get used to it. People from competition start implementing and eventually releasing competitive projects.

Sun releases, but not in atmosphere of technological progress, but in atmosphere of lacking behind their promises. People are not expecting “that cool new Sun xVM server”, they are more like “Sun again moved the release date”.

Another problem with these days Sun is something my friend Xyzz (a huge Fedora fan) described as “lack of skill in building communities”. I believe that Sun built a lot of communities and that they created lots of open-source software. But what they are doing these days is basically packaging something they can release to people in a tarball or some kind of versioning system and putting it away, so “people can play”.

But the problem is, that people don’t usually play with the source code. For example SAM-QFS release. There is some source code. No one knows how to build it, not even patch it or use it.

I think Sun is trying to sell the packaging and documentation. That may work. But it won’t build communities. Because they start with other way around. They have something, that they can use right away. They can build and try it with little time, after reading documentation, downloading a tarball or binary package and playing with it. Then they can contribute something back. Because they see it fits their needs, it just needs this little tweaking here, another feature that would be cool there, etc.

But people don’t usually get unknown untested source code just to play with it and contribute back. That is why Sun is not successful at building communities around their projects.


Written by Juraj Bednár //