[maxinc]

exactly my plan - a 64 GB SSD for the WIN7 and 2 times 2TB for storage - the question is, can i format and reinstall win7 on the SSD and it recognizes the other two disks still as in a Raid1 or how will the situation look like once i have to reinstall win7?

I haven't experienced this particular scenario but I would expect the new Windows 7 installation to recognise the raid just fine.

I've used RAID 1 many times before but the one provided by the controller. Even after controller failure, I was able to read the drives individually on different computers without much trouble. Can't say the same for RAID 0 configuration which were totally dependent on the controller they used.

[SwissElite]

ok, lets just hope it never happens, otherwise ill report here how it ended

[maxinc]

ok, lets just hope it never happens, otherwise ill report here how it ended

There is nothing stopping you from reinstalling Windows 7 (simulate crash) before you start adding data onto the software RAID1. Would probably take an extra hour but it may be worth the piece of mind

[SwissElite]

true

[Loto_Bak]

The entire machines are run off the network, no storage needed

I use PXE to network boot and then the machines mount an iSCSI drive and boot the OS.
iSCSI is basically scsi over ethernet

[BLKMGK]

Yeah, write speeds for the unRAID aren't super fast, I'm told CPU isn't the issue so I'm unsure exactly why. But I do still use it for backups and storage of ISO for Linux, torrents, etc. Programs and working files remain on local storage. Read speeds used to be lower than they are but awhile back SAMBA revved and speeds went through the roof! PXE booting my XBMC boxes would be pretty awesome but I've never been able to get it figured out I'm afraid. One day I may take the plunge with ZFS, it certainly intrigues me, but I'm currently very happy with unRAID.

FWIW my Win7 system boots from an SSD (fast!!) and I use two 1TB drives for everything else in RAID 1 I believe. I'm using the mobo RAID to do this so it's a little dicey but my attitude is that I have backups on my unRAID and nothing on here is priceless. I would be interested in hearing what folks are backing up with in Win7 - I use Acronis 2010 and it fails to run automatically on Win7 due to a permissions issue. I'm loathe to upgrade to 2011 just yet although I've used Acronis for quite a few years. This issue has been ticking me off pretty good.

Last but not least - there's a VM for unRAID posted in their forums if anyone wanted to just poke at it. I've got it but not played with it and it's only a 3 drive basic system but IMO that's kinda' neat.

[Loto_Bak]

Ive heard of people running windows home server in VM for the backup functionality. Just use your server as the storage pool.
Not a bad idea, but I havn't tried it myself

[ashlar]

Yeah, write speeds for the unRAID aren't super fast, I'm told CPU isn't the issue so I'm unsure exactly why.

every write needs to read from all volumes, otherwise how could parity be calculated? That's one of the reasons why I don't mind the snapshot approach of FlexRAID (although on the fly functionality is coming to it).

So it's not just CPU, but the disk controllers. With dedicated PCI-E controllers you get a boost in performance, but then you have to hope they are supported by unRaid.

[maxinc]

every write needs to read from all volumes, otherwise how could parity be calculated?

This is inaccurate. Only the data disk written to and the parity drive are involved during a write. The other drives don't even need to spin up.

Both data and parity drives need to read the sector written to, compute the parity based on new data being written and then write the result.

This involves 2 reads and 2 writes with a full platter rotation between the read and the write and this is what is causing the slow speed. The slower of the 2 drives determines the actual write speed.

In contrast, reads only require 1 operation and they are not affected by the parity process.

[ashlar]

Ok, thanks for the clarification (thinking about it, it definitely makes sense). In any case several I/O operations are involved which makes the CPU only one part of the equation.

And explains why a faster controller can make a lot of difference.

[maxinc]

The bottleneck is in the rotational speed of the hard drive. 10.000 rpm will have higher write speeds compared to a 5400rpm since il will take a shorter amount of time between the read and write steps of the parity-write process.

At these speeds, CPU and controllers are pretty much idle, except for the old PCI controllers with more that 2 drives connected which inherit the PCI bus bandwidth limitations.