Unfortunately no one can be told what fun_plug is - you have to see it for yourself.
You are not logged in.
Pages: 1
Hello,
First time poster, just received my DNS323 ![]()
I've been trying to read as much as possible from here and understanding good backup procedures. I believe my computer HDDs are past their service life and so this seemed like a good solution till I can afford a crazy hardware raid solution ![]()
I understand the difference between backup vs disk redundancy and availability.. and I was originally going to have 2 disks in a RAID mirror configuration, so that each disk had the data at any given time (other than when syncing..). since, and amongst other things, though now I'm wondering if for my needs would be a set up with "2 individual disks" and an rsync job running to effectively mirror them.
My reasons for this, which may be incorrect, are that there are some possible complications with RAID1 set ups in the DNS323 that can result in a loss of data in event of a single disk failure. I also understand, possibly incorrectly, that mirroring them manually does not have any specific complications yet achieves the same level of disk redundancy. Are these assumptions correct?
In the case of a single disk failure would it be easier to recover with an individual surviving disk or RAID1 surviving disk?
OTOH - I do have a gigabit network so I would like the read performance of RAID1 over 2 individual disks.. and the intention was to have a redundant backup.. so that's +1 for RAID1. Availability is less of a concern as it will be a backup.. also I don't think the DNS323 would be much faster reading to justify the complications... I do not require snapshots or restore to a particular time.. I've never had that and haven't needed it in a way the DNS323 could address.. just trying to ward off hardware failure related data loss.. arrrghh competing assumptions everywhere!
For the record I'll be putting in a WD1000EACS 1tb green power drive into it on Monday. In 2 weeks I'll be getting a second one of these drives from a different batch. Also I should mention that the power does cut out here sometimes.. I do not yet have an UPS and so the set up would need to take that into account and I'm unsure if the RAID1 would not cope with this? At work we have batteries for our RAID controllers...
Sorry for the long post - but on spending many many hours reading this great forum I wanted to check for understanding before proceeding - perhaps someone can say "no RAID1 will be easier and is safer in this case" or someone will say "individual disks are better because you don't have both disks yet and has no disadvantages over RAID1 in this case" etc..
Thanks for your time - you can see why I posted this is because with all that I've read, I'm still a bit conflicted and I'm hoping someone can mention something I haven't thought of... ![]()
Last edited by someonestolecc (2008-11-07 18:29:15)
Offline
I would just like to point out that the term "disk mirroring" implies simultaneous writes to both disks in a mirrored pair - this does not happen when rsync (or any other scheduled synchronized copy method) is used. It is only within the linux world that I have seen "mirroring" used to describe scheduled synchronization - this is NOT mirroring, this is nothing more than a scheduled backup.
The danger of scheduled synchronization as compared to true mirroring is ...
- 1 - if the primary drive fails, access to the data is lost and manual intervention is required to restore access.
- 2 - should the primary drive fail between the periodic synchronization points, any data written since the last synchronization will be lost.
Neither of these occurs with true mirroring.
The danger of true mirroring is that any deletion or corruption of data that occurs, affects both drives simultaneously - and this cannot be fixed without a backup.
With regards the dangers of true mirroring with the DNS-323 - whatever problems occur, do so at the time the failed disk is replaced and the danger can be ameliorated by simply backing up the data before replacing the drive which should be considered the first step in replacing a failed drive whether or not it's on a DNS-323
Offline
As far as performance... even with Gigabit network, I doubt you'll notice any speed improvement accessing the DNS in a RAID1 config vs. Individual Disks. A single 7200 RPM SATA disk can deliver data at speeds well in excess of what can be passed over 1 Gig Ethernet via TCP/IP, given the CPU horsepower of the DNS. So don't let that drive your configuration decisions.
That said, I am so far having good luck with my DNS configured RAID1 with a pair of WD Caviar Black 1 TB disks. I have not experienced any drive failures, nor have I tried to simulate a failure to see if the RAID1 functionality is working properly. I may do that, though, after I back up all the Volume_1 data to tape.
I do recommend getting a small UPS to support it, though. If you implement ffp with fan control and set disks to spin down when inactive, the unit draws very little power and can sustain fairly lengthy power outages. A UPS also gives good power regulation/filtration and thereby reduces stress on the DNS when power is on.
FWIW, I have a medium sized UPS and keep my Cable Modem, Wireless Router, 8-port Gig Ethernet Switch and DNS-323 all powered off of it. That way, I have a fully functional network for some time, even when we're without A/C power. I can still use my Laptop and have full internet and file access until batteries are drained.
Last edited by jesbo (2008-11-07 19:55:31)
Offline
Thanks gents - you have re-affirmed what I have read and also that it will be okay to proceed as originally planned in a RAID1 set up.
In the meantime a mate of mine has also said to mount my surviving RAID1 set up is quite easy to do with mdadm into my own pc etc.. so I can get that prior to putting a new drive into the DNS-323.
Much appreciated!
Offline
Pages: 1