11 comments

[ 2.8 ms ] story [ 20.8 ms ] thread
That's some impressive piece of software. I wonder if we are perceiving an acceleration on the rate of improvement for open source OSs. What other OSs have comparable technology?

And, more interesting, is it conceivable that FLOSS could be approaching a singularity?

The Linux kernel seems to be approaching a "singularity" in the sense of a black hole. It keeps sucking in more and more, growing exponentially larger, until eventually it becomes so big nobody can handle it or even get near it.
The announcement doesn't state what DRBD is. It turns out that it is effectively RAID-1 running over the network. It looks fairly impressive and will make setting up mirrored servers easy as cake.

Personally though I'd like to see networked RAID-0 support in the kernel. It'd be neat just throwing a bunch of machines on a network and have them all act like a giant disk.

> Personally though I'd like to see networked RAID-0 support in the kernel

No, I don't think you would. File systems won't play well with arbitrarily large drives.

You probably want a distributed file system, something like AFS, and it's already in there. http://en.wikipedia.org/wiki/Andrew_File_System

You can effectively do that now with NBD and md.
"I'd like to see networked RAID-0 support in the kernel"

Isn't that what iSCSI is for?

easy as pie.

or piece of cake.

you pick. :)

The idioms you're familiar with and the idioms the parent is familiar with may not be identical. I've heard "easy as cake" used reasonably widely.
Distributed Replicated Block Device (http://www.drbd.org/) can be used in conjunction with something like Linux HA (http://www.linux-ha.org/) to create a systematic high availability clustering approach for stateful (and transactional) applications such as a database. Lots of sites use it (e.g. github.com IIRC) and up until now, you've had to run a patched kernel to take advantage of it.
Use AoE and you can have whatever raiding system you want.
See Phillip's response for a bit more explanation in the mail thread:

http://marc.info/?l=linux-kernel&m=125309001007725&w...

At my (soon-to-be-last) company, emarsys, Linbit maintained our production clusters. We had a couple of Postgres servers and main file storage replicated across physical locations - a customer requirement, actually, that the system be capable of handling something like a fire at one of the data centers.

Overall, it worked well. In the 2.5 years I was there, the only time I needed a backup was to restore my own mistake. :)

I'd check it out if you're looking for a more complete replication system (Again, not RAID.)