Matthew Dillon’s been thinking about Hammer, and how to implement clustering well enough to work as a sort of RAID replacement. He’s written up a document describing his plans. Some highlights:
- writable history snapshots
- quotas and accounting
- live rebuilds of data from mirrors
- and the same history, mirroring, and snapshots as before.
It’s going to be a while before this “Hammer 2” becomes a finished product, though, so don’t count on it for the next release.
1) A little disappointed that HAMMER is already obsolete and needs a “complete rewrite”
2) Quotas: I thought the goal was to implement it in the VFS layer?
Obsolete would imply that all those features are no longer useful. It’s going to work the same as before with snapshots, mirroring, history, etc, but he’s changing the underlying data format to handle multi-master work better. That’s all.
I’m really interested in using the mirror capability; I think even using it locally could let you do some neat things with SSDs or even with tmpfs systems…
Maybe “obsolete” wasn’t the right word – more like – “end of life”.
End of life is what things like Windows XP are experiencing. A better phrase would be “developing”.
“HAMMER2 has no ‘atime’ support… The OS is expected to implement a devfs for the device space (the only space where atime is really used).
Hm… it used to be that atime was used to detect new mail in mboxes. Has everyone switched to maildir or webmail now?
A while ago, Linux switched from atime to relatime. It seems a good idea — all the benefits of atime, but significantly faster.