Benchmarks Of ZFS-FUSE On Linux Against EXT4, Btrfs

[smitty3268]

- instaneous recovery (zfs rollback)

Well, after you fix the state by rolling back any open transactions. I'll give you that it's faster, though.

- instaneous database cloning (useful when doing upgrade simulation, additionally, it doesn't use extra space until new data is written)

Now that does sound useful, although only in very limited scenarios - i.e. testing stuff on my local machine or during upgrades.

- configuration-less backup

Well, you've got to configure a timed job somewhere, unless you're advocating backups on the fly whenever you happen to remember to want one...

- centralized, simple backup of everything, not just database (useful for all-in-one servers)

Well, that's true, i suppose. But it's more of a crutch to allow you to do things poorly if you don't have the resources to properly back things up, if you ask me.

Well, if those above don't matter to you, yes, probably there's no benefit in using zfs snapshot

There is actually 1 time when I used a VMWare snapshot (which is very similar, only it backs up the entire machine state with the disk, and i'm sure is much slower) and that was when going through a big database upgrade. It was nice to know that if everything went to ****, i could just touch a button and restore the whole machine back to working order without having to create a clean install, restoring all the data, etc.

Very comforting to know that option existed, although I can't see using it for anything else.

[edogawaconan]

And again, I ask: What's the point?

I've got transactional logs being created every 10 minutes and getting backed up. What would this possibly gain me in real life and not just theoretically?

Just because something CAN be done, doesn't mean it SHOULD be.

- instaneous recovery (zfs rollback)
- instaneous database cloning (useful when doing upgrade simulation, additionally, it doesn't use extra space until new data is written)
- configuration-less backup
- centralized, simple backup of everything, not just database (useful for all-in-one servers)

Well, if those above don't matter to you, yes, probably there's no benefit in using zfs snapshot

[smitty3268]

By at all, I mean no delayed datafile writing, no additional write, and done in an instant.

And again, I ask: What's the point?

I've got transactional logs being created every 10 minutes and getting backed up. What would this possibly gain me in real life and not just theoretically?

Just because something CAN be done, doesn't mean it SHOULD be.

[edogawaconan]

As can transactional logs, so really I fail to see what this gains you in the real world and not just theoretically.

By at all, I mean no delayed datafile writing, no additional write, and done in an instant.

[smitty3268]

All this can be done without disrupting database process at all.

As can transactional logs, so really I fail to see what this gains you in the real world and not just theoretically.

[edogawaconan]

It is poor data management. Transaction logs shoulb be the last line of defense against failure, not the first. Timely, application specific backups should always be your first line of defense. ZFS snapshots, in this case, depend on the database engine's emergency recover processes as your only line of defense.

You went from "not a backup" to "only line of defense".
Any database worth using should have no problem replaying transaction log.

Also, creating zfs snapshot only takes one command and completes under a second. No need to do additional configuration and the snapshot can be sent incrementally to other machines. All this can be done without disrupting database process at all.

[Kano]

ntfs-3g uses fuse too. maybe also interesting to compare.

[dacresbu]

FUSE Performance

Typo, yeah, meant FUSE. Fixed. Thanks.

ITS IN USERSPACE! If you asked for performance, your missing the point of FUSE. The point is to do things normally too unstable to support in kernel space. Doing this in userspace is of cores slow. As for getting this to near platter speed, that is impressive.

[locovaca]

Powering down computer with switch has risk of corrupting partially written data. It doesn't happen in zfs snapshot. How is it exactly same?

It is poor data management. Transaction logs shoulb be the last line of defense against failure, not the first. Timely, application specific backups should always be your first line of defense. ZFS snapshots, in this case, depend on the database engine's emergency recover processes as your only line of defense.

[edogawaconan]

As I said, it's the exact same as powering down your computer with the switch.

Powering down computer with switch has risk of corrupting partially written data. It doesn't happen in zfs snapshot. How is it exactly same?