DSM-G600, DNS-3xx and NSA-220 Hack Forum

If the DNS-323 dies, you should be able to remove the disk from the enclosure, install it in a PC running linux and retrieve your data - you can also do this from a Windows system using an installable file system driver such as EXT2 IFS (http://www.fs-driver.org/).  Ideally you would also have backed up your data elsewhere, since RAID is not a form of backup.

Why should you use RAID if all the data will be backed up anyway?

Simple - the whole purpose of RAID(1) is to reduce the impact of the downtime caused by a disk failure.

Think about this .....

Let's assume you don't use RAID, you backup all your data, and at 9:00 am, your disk fails - you have no access to your data until you get a replacement disk, install it, and then restore your backup - how long does that take?  That's the down time.  Now how much is it costing you per minute/hour?  That's the impact of the down time.

Let's assume you use RAID(1) - at 9:00 am, your disk fails - you still have access to your data, there is no down time, there is no impact.

And since you're going to ask why is RAID not a form of backup, let me answer that - what happens if a user deletes a file, whether by accident or deliberately?  What happens if a virus corrupts your database?  With RAID(1), both copies are affected immediately, your data is history - with a backup, you might lose some data (depends on when the last backup was done) but you should have the bulk of the data intact.

BTW - this question has been asked before - if you search you'll find the solution I've suggested - along with other suggestions

Edit - As Haydn kindly pointed out, I slipped up with the wording of the first sentence, it has been corrected.

Last edited by fordem (2008-03-14 18:18:25)

Thanks much for that Haydn - and I'm working hard on that disk resurrecting computer - Bill Gates will have to step down as the world's (what is it now) third richest man now once I get the bugs worked out.

timofej

Search the forum, there's a thread on an automated backup from disk1 to disk2 that might meet most of your back up requirements - and if you don't map a drive from the PC to that second volume, you should be safe from the effects of a virus.

Yes, you should be able to find external enclosures that allow SATA drives to be connected to a USB port, in fact I use a little SATA/PATA to USB adapter, without any enclosure - something similar to this - http://www.apricorn.com/product_detail. … &id=39

I agree. I think RAID (both 0 and 1) is overrated on a DNS-323.

Thank you guys for your answers! They are really helping me to make the right decision.

Since the risk of the down time is not critical for my system RAID1 seems not have any benefit for me. Both fordem and mealto have a point that data corruption of any reason (unit fail, virus attack etc.) will lead to complete data loss. Automated periodical back up from disk 1 to disk 2 will loose only limited the data amount (if any at all).

I have never considered this type of back up. I was completely hung up on having another server which would yield additional costs and connections. Considering the possibility to connect SATA disks to a PC the idea of backing up to disk 2 is just great!

Thanks again for your help! I think the issue is solved. Will do some research how to make back up between the disks now. :-)

Given the size of the DNS-323 it could be tucked away somewhere out of sight and/or secure and the PCs backed up to Vol1 and then Vol1 backed up to Vol2 - here's the tutorial on that.

This method has two advantages over backing up the DNS-323 to a disk in a PC - the first is that there is no traffic across the network, and the second is the issue of linux/DOS-Windows archive bit mapping, which more often than not causes problems with backup software.

fordem wrote:

Given the size of the DNS-323 it could be tucked away somewhere out of sight and/or secure and the PCs backed up to Vol1 and then Vol1 backed up to Vol2 - here's the tutorial on that.

Thank you fordem for the direct link. I don't need to search myself now. :-) Will report the results here when I'm done. Even if it can take a while - don't have so much time to play with my equipment.

This forum is great! As well as the DNS-323 itself since it is possible to hack. I have browsed the forum for a while and is amazed how customizable the device is using the fun_plug. I wonder how come that this particular device has become almost iconic among NAS users. It is surely 3 years old but is still extremely popular. All other electronics are considered to become crap after a couple of years but not DNS-323. There are plenty Unix based NASes there but I have not met such dedicated society around any of them as I've found here. But this is just a little offtopic. :-)

fordem wrote:

I could be wrong - but I think it's just over a year old, I can remember hunting online for mine, when it was first released, and I'm thinking fifteen months or so.

Yes, your are right.:-) The device was announced autumn 2006. I don't know why I decided that it was 3 years old.:-O :-) However it's still a long time for electronics. In Sweden where I come from it is the most popular product among NASes. If you look at other categories the most popular products are seldom older than 5-6 months, that would mean that the DNS-323 is a really old peace of electronics. :-)

P.S. Ok, it's enough for today. It's getting quite late here. Have to go to bed now. :-) See you around...

Hi Blahsome,

blahsome wrote:

For a home/SOHO device such as the DNS-323, you'd much more likely experience an accidental overwrite or deletion or corruption or power interruption than hard disk failure. None of these scenarios benefit one bit from RAID1.
...
Sorry, I still fail to see any benefit of either RAID0 or RAID1 *on a DNS-323*.

Two points:

1/  I do see benefits of RAID1 in relation to the power interruption scenario.  This happens frequently in SOHO environments.  The DNS-323 is prone to disk corruption on abnormal power interruption.  I must add this is not a defect but a property of the operating system that is running on it.

In two instances at at least, I was able to recover their systems easily and remotely because the DNS-323's were running RAID1.  The process would have been much more difficult if RAID1 was not running.

2/  Also, in a SOHO environment, the DNS-323 (or like devices) on a UPS running RAID1 is a nice solution for a remotely managed storage using with near 100% availability using inexpensive disks.

My 0.2cents worth

Jaya

blahsome wrote:

I do understand the benefit, and your assumption is wrong.

First, note that I qualified my statement with "on a DNS-323." If you use a DNS-323 as part of a mission critical system, all power to you. For a home/SOHO device such as the DNS-323, you'd much more likely experience an accidental overwrite or deletion or corruption or power interruption than hard disk failure. None of these scenarios benefit one bit from RAID1.

Second, somebody has already argued that RAID1 on a DNS-323 is not supposed to be your primary backup method; I won't repeat it here. Under that assumption, any of your important data is backed up somewhere else. Why then should I waste half of my storage capacity on yet another copy?

Sorry, I still fail to see any benefit of either RAID0 or RAID1 *on a DNS-323*.

There are many small businesses whose owners would rather be running their business than managing servers, they don't want and in some cases can't afford a true server solution, but that doesn't mean that they don't need reliable networked storage.

The SOHO market was D-Link's target for the DNS-323 - in many cases the initial setup will be done by a "consultant" or tech savvy friend and all the business owner has to do is power it up in the morning with a touch of a button, and back off in the evening, using the same button.

As regard the RAID1 not being primary backup, there's a good chance that "somebody" was me - go back and read the post, and you'll see why a small business owner should waste half of his/her storage capacity on yet another copy - because - if/when a drive fails, he/she can continue to work rather than to have to wait for the consultant or tech savvy friend to come and sort out the problem.

Not having immediate access to your data might not be a problem for you, but it frequently is for some of us - RAID1 *on a DNS-323* might not offer you any benefit, but it does for many others.

Hi Blahsome

blahsome wrote:

A simple RAID1 implementation guards against *mechanical disk failures*, nothing more.

Firstly I don't know what you mean by "simple" RAID1.  Secondly, it is not just mechanical failures that RAID defends against but electrical and other failures too.  Most importantly "nothing more" at the end of the your statement IMHO makes it absolutely incorrect.  You don't know what events RAID1 implementations defend against unless you look into the implementation.

How do you know the I/O state of the two disks when a power failure occurs? You don't.

You do if you look at the implementation.  In my experience I found that software RAID1 implementations tend to have less concurrency than their hardware counterparts.  This makes them more robust in relation to power failure events.

RAID1 "mirrors" data on both drives, meaning a power failure can well corrupt both drives.

The mirroring policy determines if power failure will corrupt both drives.  If in a software RAID implementation, the policy is to write to only one drive of a mirror set at any one time, then corruption of more than one drive as a result of power failure is extremely unlikely.

Let me quote from the following documentation: The redundancy of RAID levels is designed to protect against a disk failure, not against a power failure.

Nothing wrong with the statement in the documentation, but your inference does not seem to follow.  RAID is not designed to protect against power failure.  That does not mean RAID implementations will not protect against such an event.  You have to look at implementation rather than rely on one liners.

Regarding high availability, I've seen reports where the DNS-323 got confused and built or rebuilt the RAID arrays incorrectly. The moral of the story is that you are NOT supposed to store the only copy of any important data on a consumer device such as the DNS-323. You need to back it up somewhere else too!

I read the anecdotal evidence differently.  All the reports with RAID1 point to availability of data even when one of the drives shows up with amber light.   All reports with JBOD  under these circumstances all lost data.  So the moral as I see it is quite different to yours

If that's the case, again why should I waste half of my storage capacity on a RAID1?

By definition, RAID1 trades 50 percent space to defend against single drive failures.  You don't have to use it if you do not need the functionality it provides.

Jaya

jayas wrote:

...The DNS-323 is prone to disk corruption on abnormal power interruption. 
I must add this is not a defect but a property of the operating system that is running on it....

@jayas - Disk corruption (on power outage) is more a property of the non-journaled file system (ext2)
D-Link choose to use.  With a journaled file system (ext3, xfs, riser) the probability of disk corruption
on abnormal power outage is greatly reduced.

@blathsome - I use RAID1 to keep my data available in the event of a hard drive failure.  IMHO
the additional expense is an acceptable premium for this type of insurance policy.   YMMV

Last edited by mig (2008-03-15 06:30:03)