• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Help choosing a backup solution

M

mikegeig

n00b
Joined
Jun 30, 2010
Messages
11
Hello everyone, I am very new to Disaster recovery and backup solutions, so any help would be appreciated. I am purchasing (2) Servers to run Hyper-V. In total I plan on have 8 virtual servers running across the 2 machines. Each server will have a raid 5 configured with (3) 600gb hard drives. Realistically though, I plan on having 500gb full backups with 100gb dailies. I will also have a 3rd server who's role is simply to backup the 2 virtual servers to media.

So my question is what solution should I go with? There are many options like tape, hard drive, cloud, and I am having trouble figuring out what would be best suited to my needs.

My budget is approximately $4K, and any help would be greatly appreciated.

Thank you
 
I would normally suggest LTO tape since I do not trust disk as a backup however you have a small amount of data so you can make multiple copies with disk. The cost of a LTO4 or LTO5 drive would not make as much sense with only a few tapes.

In your case I would get at minimum 7 1TB external drives with esata and swap them out daily. Also add a few extra for redundant full backups.
 
why not just build two identical servers and have one mirror the other, have a RAID5 with six 600gb drives (maybe a hot spare too), then depending on the filesystem and operating system you choose- just partition them up for the different virtual machines.

in your setup, if one server went down you would have the data backed up but would have to manually get the main server back up and restore it to functionality. meanwhile the users are offline. (unless the two primary servers you mentioned were going to be mirrored).

in my suggestion, if one went down, you could configure it so that the end user wouldnt even notice- service would just fall back to the other server and you could have time to restore the primary server without any down time. this also enables the possibility of load balancing between them if one ever got overwhelmed.

it might help to know what type of environment these will be running in. is this a personal network? something for a business/your job? number of users? file serving? web hosting? it all depends on what your data and uptime is worth to you/your clients.
 
Nope. One example how this is not a backup. If a user accidentally wiped out some of their irreplaceable data without knowing and this change got automatically mirrored to the second server (before the user found out) you would have no copy of the accidental deleted data.

With a backup you would have multiple copies to fall back on.
 
Clustering like RAID is not backup, but rather is just uptime.

There is no "best" solution, or "right" solution. It all depends on your situation.

If you have a decent connection to the internet, and your data set doesn't change vastly each day and you can afford the monthly cost, cloud backup can be a very good solution. (Mozy Pro, iDrive, Carbonite etc) but it is slow. However it easily takes care of off site, etc.

Tape is the old standby (and what I am using currently along with some disk). There are lots of types of tapes, LTO being the standard, AIT being worth a look etc.

To disk is fast, but if you use it exclusively be aware disks are not as robust as tapes in most cases.

Usually some mix is the best solution. I use tapes as my fall back - worst case solution, but I also do Volume Shadow Copies that have actually been great for quick file restores. And my Exchange doesn't actually remove things from it's database for 30 days, so I only have to go to tape if something OLD is missing. And I also backup my SQL databases both to a local disk and to tape, that way everything I have is backed up both on disk (for speed) and tape for redundant and off site. I also do some DFS to remote locations and disk backups of the data there.
 
ghost6303 - a reasonable backup should also protect against theft and fire/water (fire sprinklers most likely). But also thinking about uptime is good.

My wife's company had to send everyone home (160 employees!) around 2pm one day last week because a server went down. The next morning they could do their work but didn't have e-mail until around lunchtime. The 3 hours of downtime was probably about $10,000 in wages alone, or something like a $35,000 loss to the bottom line that day.

At my work we're small potatoes with 6 users, but if our primary server went down I can switch to the other one with only about 10 minutes of downtime. We also rotate LTO-2 tapes off-site daily.
 
I am looking at a RD1000 drive with 640gb hard disk cartridges. It seems to be a fairly cheap and effective solution. What do you think? Have you heard anything bad about them?

Ghost6303: I didn't know you could do that. What is that actually called (technical name) so I can research it a bit. How does that affect storage? Instead of having 2.4 tb total would I just have 1.2 since they are mirrored?
 
I see.

Will your data set grow considerably over time? The reason I ask is additional cartridges are very expensive in my mind well at least compared to tape where a cartridge (LTO2, LTO3,LTO4) less than $50.
 
I don't believe I will see rapid data growth. As it stands, I can buy the RD1000 drive for $389. An LTO 4 drive would cost me ~$3500. Even at a $100 difference in media, the RD1000 is still much cheaper.
 
I was thinking LTO2 or LTO3. These can be had for less than $1500 for a drive although that would require manual work since LTO2 is 200GB native and LTO3 is 400GB native. An autochanger would remove a lot of the manual work however that would not easily fit in your price range especially if you need to purchase backup software for that. If your dataset does not grow your solution is fine.

If I was in your situation I would have just done the backups with a dozen or so ~$100 1TB or 1.5TB esata mybooks or similar. Although I am always trying to keep cost down even at work. With that I would have used good backup practices like Grandfather-father-son.

http://en.wikipedia.org/wiki/Grandfather-father-son_backup
 
Last edited:
Even after reading the link you posted, I don't quite understand it. Why would you do something like a gandfather-father-son system instead of just taking daily full backups?
 
In a good backup procedure you always want multiple full backups. This is just one way of rotating the media to accomplish this.
 
What is the benefit to this versus just doing daily fulls though? It seems like added complexity for no gain, unless I am reading something wrong.
 
Daily fulls will be fine if during the backup you do not use the previous days media.

However Grandfather-father-son is still better however because if some files were anciently deleted last week and you have already overwritten all backup media from then you have no backup of the missing data. This would also happen with the rotation I mentioned if you extend this to a longer period.

At work I reuse media after 1 year of retention. I am not using this rotation strategy however. On user folders and systems I run daily incremental, monthly fulls, weekly differentials but the backup program manages the media with the retention periods I specified for each pool.


BTW, like the Lazn_Work said there is no best way that fits all. It all depends on your parameters.
 
Last edited:
Ah, I see. Here we have a 1 week retention. If a user deletes a file and waits a week to tell me, then it is gone. With a year of retention, you must have a closet full of media.
 
You must log in or register to reply here.
Back
Top