Andre Nathan asked for links on SSI (Single System Image), since that’s DragonFly’s pie-in-the-sky goal. Walt gave a link to Larry McVoy’s paper (Linux-specific) on the subject.
Disclaimer: what follows is my opinion, nothing more.
One of the things that I like about DF is the fact that it is being redesigned at it’s core to enable such features as this natively, as well as to enable future additions without needing to do another core redesign. From where I’m standing, the DF folks are basically future-proofing their OS (as far as this is realistically possible – nothing lasts forever).
Although I do understand the (for example) Linux developers desire to not make far reaching alterations to their kernel while adding new features (such as clustering capability), in my mind it ultimately leads to a system that not only looks cobbled together, but is more complex than it really needs to be in order to maintain all of the accumulated features.
That said, it cannot be claimed that the Linux folks haven’t done a fantastic job of extending their kernel, as it is currently the most scalable free OS available.
If you’re interested in SSI, then be sure to have a look at http://www.openssi.org — it’s Linux based, but it’s already pretty complete. A V1.0 release is on the near horizon.
Work is on the way to include http://www.drbd.org/ so that you can have a HA cluster without shared disk hardware.
Don’t know if any of it would be useful for DragonFly (probably not directly, but being able to look at someone elses solution is always useful, no?)
I doubt that it would be useful to DragonFly in and of itself. DragonFly is being redesigned at its core to allow clustering capability, and so it won’t really need third party stuff to give it clustering capability.
That said, it may well be useful as a reference for designing a compatibility layer that would allow DragonFly’s upcoming technology to talk with HA clusters.
DragonFly BSD Digest