Setting up a fileserver - help and backup method/recommendations

[Completely Bonkers]

I'm setting up a basic Windows fileserver to replace a Linkstation NAS. It is a relatively low use/demand, with not more than 50 new or editted files per day, not more than 50MB of changed, and 5MB of new material per day, on average.

The main HDD (640GB) is as follows:

C:\ system partition 20GB
D:\ data drive, requiring frequent backup 300GB
E:\ data drive, requiring occasional backup 300GB

I'm looking for advice/suggestions on a suitable data security / backup / recovery method.

My initial thoughts are the following:

1/ Set up Volume Shadow Services to maintain a daily file-copy on the main HDD, which provides rollback to the user in the case of basic file management mistakes
2/ Install a second HDD (also 640GB) on the machine
3/ Run Cobian backup to make a copy of the D and E drives on to the second HDD. There would be a full backup once per month, plus a differential backup done daily at 2am.

Is this a good setup? Any other ideas? Suggestions welcome and appreciated.

I have also seen Acronis Backup & Recovery, but this is too expensive. Looking for cheap but effective solutions. Can spend to a small budget, but not hundreds.

 

[CDdude55]

Buy a second 640GB HDD and put them in a RAID 1.

Im not an expert at the best backup methods, but that's what i would do if i wanted a constant backup of data and fault tolerance.

 

[newtekie1]

Just use Robocopy(built into Windows Vista and 7, downloadable for XP).

Write a bat file and schedule it with task scheduler to back up the data to a second drive as often as you want.

 

[Completely Bonkers]

Buy a second 640GB HDD and put them in a RAID 1.

Im not an expert at the best backup methods, but that's what i would do if i wanted a constant backup of data and fault tolerance.

Thanks dude,

The problem with RAID1 (and why I didnt go for it) is that is provides fault-tolerance for hardware failure* on one of the drives. So if one HDD goes bad, the other HDD maintains a working system until the first one is replaced. So it is good in a 24/7 system to keep the system live. And data isnt lost due to one HDD failing.

The downside is that there is still no backup. So if there is a NTFS/MBR/MFT failure, this is duplicated on the other drive. ie. it doesnt protect against "software failure". There is also no protection if someone carelessly deletes or edits a file. The original data is lost.

That's why I went down the daily backup route. I'm more interested in collective data protection than trying to maintain a 24/7 environment against hardware failure.

In the best of worlds, I would do RAID1 and then still have a backup system, and probably that backup system should be in a second physical location. Food for thought.

 

[Completely Bonkers]

Just use Robocopy(built into Windows Vista and 7, downloadable for XP).

Write a bat file and schedule it with task scheduler to back up the data to a second drive as often as you want.

Thanks newtekie

Took a look. Seems good. I think Cobian Backup is essentially doing this with the advantages of being GUI based and also can work with VSS (volume shadow services) so that is can also backup "open" files in case a PC has crashed or stayed on overnight with open files.

 

[Completely Bonkers]

I'm currently thinking through the issue of "remote management" if the server is in a different physical location that the admin, and what levels of backup/security/restore might be needed in all events.

>> Question. In RAID 1, if a drive fails, how long can it be left in the array as failed, or does the array require immediate replacement and rebuild? Because if I no-one is physically there for a month or two...

 

[newtekie1]

If I drive fails there really is no limit to how long it can be left there. The controller will just not use it, and function as a single drive configuration, waiting for a replacement.

The only downside to this, is that sometime when a hard drive fails, it will still sort of work(I'm sure we've all seen that, when a hard drive is on its last legs and starts acting flakey). So the controller can mark the drive as failed because it started acting flakey. But then if the machine is powered down or rebooted, the controller can see the drive as being there again and start rebuilding the array with it. Of course this isn't too bad because the rebuild will usually just fail and the drive will be marked as bad again, but it does waste some resources during the rebuild process and depending on the controller can eat up a lot of CPU time with a rebuild that is worthless.

 

[bpgt64]

Completely agree with Newtekie1, Raid 1 is what you need. Since your not backing up a larger files(media/music), I would go with a second 640GB drive and a Professional copy of Windows, for added remoting capability. For simple access to your files offsite, you can either setup FTP software, or checkout Dropbox or a similar service.

 

[Completely Bonkers]

newteckie, thanks for the RAID1 info. So if a drive fails the system doesnt insist on a replacement HDD to rebuild the RAID. Good. It means that if I get onsite only once every 3 months, a problem can sit there waiting to be fixed.

bpg, thanks for suggestions. As per my earlier reply, RAID1 is no backup. It is only a hardware fault-tolerant setup. Useful, but not my primary concern. Think deleted file, think virus, think idiot co-worker who deletes or overwrites something important.

I WILL HOWEVER think about getting another HDD to set up RAID1 in addition to a backup procedure.

I should add that I want to set up a physical-maintenance-free remote office. I will be there only once every 3 months. There are no IT-competent people on the site. So if RAID1 helps in this instance it is a good idea.

A BIG THANK YOU for the introduction to drop-box. Nice features. Nice, simple, to-the-point website. (If only Microsoft could do this! LOL). However, I would prefer to run the "server-side" application on our servers rather than on a 3rd party unknown longevity provider. For business continuity and data security reasons this needs to stay in-house.

Does anyone know of a "timemachine"/dropbox type of data replication service/application for regular windows fileserver? Back in the 90's Lotus Notes used to be good at this kind of thing.

 

[bpgt64]

NP, I just went through all of this, and finally I am reaching the functionality that I want. That is, I have a 6TB(5x1.5TB Drives) in raid 5. After learning pfsense and Fedora 13 enough to setup my own FTP. Very nice for getting access to your files from anywhere

 

[Completely Bonkers]

You know, the more I think about it, the more I feel we should pony up the cash and RAID1 our fileserver and our webserver. Avoiding just ONE downtime and recovery when we have a remote site is probably reason enough. Oh, if only I had deeper pockets...

Now I need to check if these Atom ION mainboards do RAID.

Question 1: If I have an existing server using just one HDD, how do I convert it to RAID1? Do I have to reinstall windows from scratch on a RAID1 formatted system and migrate the data over? Or can I just plug it in and the RAID1 firmware will "build" the second HDD automatically?

Question 2: Can I operate a RAID1 pair and a third non-RAID drive on the same machine, same HDD controller? (ION chipset, no room for 3rd party RAID controller, these are 1U racks, squeezing in 3x 2.5" drives is already going to be tight)

>> I am beginning to go down the following thought path; 1/ Run RAID1 on the fileserver with 2x identical drives, 2/ Run RAID1 on the webserver with 2x identical drives, and 3/ set up a new server that acts ONLY as a backup server, running midnight scripts to backup the fileserver and the webserver. More hardware, more electricity, more cost, but probably "simpler" and also backup is now on physically separate hardware. Or is this complete overkill, better to have the webserver backup the fileserver, and the fileserver backup the webserver, onto 3rd drives.

Question 3: Does anyone have experience of the Seagate Momentus XT drives with the inbuilt hybrid 4GB flash? Because the utilisation of the webserver is relatively low, and to save power, we run power-management and let the HDDs spin down after 15mins inactivity. Problem is that when there is a new webhit, the server doesnt respond until the drive spins up again. This creates an ugly latency. Would the XT drives avoid that?