Suggestions / Comments for Backup Strategy

[vaicsimmons]

I am new to Veeam and I’m looking for a bit of advice. I currently have 15 VMs all running Windows Server 2012 R2 that I would like to backup. According to Veeam B&R, I have about 1.5 TB worth of VMs in total. If you consider that I am running Windows Server 2012 Deduplication on my file server, I am around 1 TB in total data between all the VMs.

Most of the data on our file server is pretty stagnant. The growth of our file server is pretty slow. Without Windows Deduplication, I have about 600 GB of storage in total on the file server. If I was to guess, there is well under 1 GB of files added per week. The other active servers would be our Exchange and SQL servers. The Exchange server has roughly 50 mailboxes on it. The SQL Server has a handful of databases with the largest being 8 GB. All-in-all, we are a small shop.

I am looking at purchasing either a Synology RS2414+ or a Synology RS815+ to house my onsite backups. My thoughts are to keep 7-14 days of restore points, a months worth of weekly backups, 12 months of monthly backups and 7 yearly backups.

If my thoughts are correct, I can run Veeam B&R in a Windows Server 2012 R2 VM. I can then add two new disks to that VM and have those disks stored on the Synology device. I can then enable Windows 2012 deduplication on the disk / volume that holds my achieved (weekly, monthly and yearly) backups to minimize the actual storage space consumed.

With all of that said, since I am new to this…I am looking for some feedback on my strategy. Do you guys / gals have any comments or suggestions on using a Synology device to store my backups? Any comments on my backup plan are appreciated as well.

Thanks in advance for all your suggestions and help!

[k00laid]

What you describe is very much so possible and what a lot of folks are doing, but it has fallen out of favor with Veeam due to Windows Server 2012 R2's handing of large files. My understanding is that files larger than 1 TB (which your backup files would be) have a high potential for corruption that can ruin an entire chain. All that said I am doing the same and haven't run into the issue in a long time. My major suggestions if you are going to go this way are the following:
1. Don't rely on a single backup system. Have two devices, even including tape, that you step down to via backup copy or tape backup.
2. Format the volume on which you are going to store backups with the /L parameter. This will tell NTFS to use large size file records allowing for greater file size.
3. In setting up your deduplication on the volume where you store backups make your dedupe window plus 1 day over your window for synthetic operations. I found that if I left the default of 3 days when I did a synthetic full on the weekends it was exponentially slower. For example my window is set to 8 days.

[vaicsimmons]

I have been playing around with Veeam and just backing up to the SAN that houses all of my production VMs. I have watched a number of webinars hosted by Veeam and read a lot of forum posts and articles. Based on my research, I have my backups set as forward incremental with inline dedup enabled. The compression level is set to Optimal and the storage optimization is set to Local Target. I am not taking synthetic backups but I am running full active backups every Sunday. If I am not mistaken, from my research it was shown that full active backups are performed quicker on a Windows Dedup volume than synthetic fulls.

[foggy]

Chris, I have a bit of reading for you, before you start implementing you strategy. I assume you'll find all of the following topics relevant to your planned backup infrastructure:

- First of all, you can search for existing topics about Synology devices.
- Please also review our general recommendations for the backup target and our reference architecture (another reason to consider some additional storage is the fact that you would need to run backup copy jobs to achieve required GFS retention and doing so within a single device is not optimal in terms of performance).
- Topic dedicated to Windows deduplication on Veeam B&R repository: Best Practice for MS Server 2012 DeDup Repo.
- Also note that storing backups on VMFS is not considered the best practice.

Hope this helps.

[vaicsimmons]

I am new to Veeam and I’m looking for a bit of advice. I currently have 15 VMs all running Windows Server 2012 R2 that I would like to backup. According to Veeam B&R, I have about 1.5 TB worth of VMs in total. If you consider that I am running Windows Server 2012 Deduplication on my file server, I am around 1 TB in total data between all the VMs.

Most of the data on our file server is pretty stagnant. The growth of our file server is pretty slow. Without Windows Deduplication, I have about 600 GB of storage in total on the file server. If I was to guess, there is well under 1 GB of files added per week. The other active servers would be our Exchange and SQL servers. The Exchange server has roughly 50 mailboxes on it. The SQL Server has a handful of databases with the largest being 8 GB. All-in-all, we are a small shop.

After posting an earlier thread noted below and doing a bit more research, it seems that I am better off running Veeam B&R on a physical server with local storage for the repository then use Veeam in a VM and have the repository on a NAS such as a Synology device. With that said, I came across an HP ProLiant DL180 G6 that holds 14 HD Trays (12 in front & 2 in back) on eBay. The auction can be found below:

http://www.ebay.com/itm/HP-ProLiant-DL1 ... 3cdf8521bb

My plan was to begin to load this up with 6TB Western Digital Red Hard Drives. This server will be used to hold 7-14 days of restore points, a months worth of weekly backups, 12 months of monthly backups and 7 yearly backups. I was thinking about putting Windows Server 2012 R2 on this server and enabling Windows Deduplication to minimize the storage requirements for my achieved data.

Do you think that this server will fit my needs? In your opinion, should I use the p410 RAID controller to setup a RAID 6 array or use JBOD (using the p410 "hack" of assigning each drive to a RAID 0 array as the controller doesn't support a JBOD mode) and utilize Windows Storage Spaces? Ideally, I would want to start off with a minimum amount of drives and add additional drives as needed.

All suggestions and comments are appreciated!

My previous thread - http://forums.veeam.com/veeam-backup-re ... 27963.html)

[lsumarkb]

Question @vaicsimmons: I have a similiar environment to you that I am quoting right now. Can you add WD Reds to an hp/dell/ibm server like that? Would you not be forced to use HP's drives?

[Shestakov]

Hello Chris,
I`ve merged back you topics since it looks more logical to revise your infrastructure in the same topic.

Regarding your questions,

Do you think that this server will fit my needs?...
... This server will be used to hold 7-14 days of restore points, a months worth of weekly backups, 12 months of monthly backups and 7 yearly backups.

What is the size of your Full and Incremental backups?
Do you have 4 backup jobs now? Do you want to keep weekly/monthly/yearly bakups as Full backups?

I would suggest to follow 3-2-1 backup rule for the best data protection. It means, keep short-term backups on the local site for the best RTO and archive historical aka long-term backups on the different storage. It can be achieved by use of backup copy job with GFS retention. It can help you to save space deduplicating monthly/weekly/yearly backup overlaps.

Thanks.

[vaicsimmons]

Shestakov,

I just started to play around with Veeam B&R. For testing purposes, I am saving all my backups onto my SAN that houses all of my VMs. It seems that my full backup had a Data Size of 2.1 TB and a Backup Size of 732 GB. My incrementals are around 15-20 GB in Backup Size. I am guessing that most of this space taken up by the incrementals is from our Exchange and SQL databases as we do not add that much to our file server on a daily basis.

I only have one job setup now for testing as I do not have the space on my SAN to store all the backups that I would like to take. I would like to keep my weekly / monthly / and yearly backups as Full backups. My thought process is to store all backups (daily incrementals, weekly fulls, monthly fulls, and yearly fulls) on a physical server stuffed full of drives that is running Veeam B&R on my local network. I will then perform a backup copy job to move the weekly, monthly and yearly fulls from our main office to storage device in another building of ours 1/4 mile down the road (connected via fiber) to get the backups out of our main office. In addition, I may also copy them to an online service as extra insurance.

[Shestakov]

Thanks for the clarification, Chris.

Your reasoning looks absolutely sane.

If you store historical backups both locally and offsite, 6TB of WD Red in the local will NOT be enough for keeping 14 (1 Full + 13 increments) restore points =~1TB + 23 more Fulls (4 weeklies + 12 monthlies + 7 yearlies) = 16.8TB, thus 18TB in total.

Simple calculation shows that besides 14 short-term restore points, your WD Red is capable to fit only 6 more Full backups. So, I`d suggest to set local weekly/monthly/yearly jobs to use increments and use backup copy job with GFS retention to store historical backups offsite once you install the offsite repository. Thanks!

[vaicsimmons]

Shestakov,

For clarification, the local storage server will not have a total of 6 TB of hard drives....it will be filled up with multiple 6TB hard drives. I have switched gears a little bit as I have done more investigating as it seems that the HP server only likes HP drives. I am now looking at the following SuperMicro Server on eBay -

http://www.ebay.com/itm/2U-Supermicro-1 ... 20fb6dca3d

It has 12 hard drive bays that I could either fill up with 4 TB Hitachi 7200 RPM Ultrastar Enterprise drives or 6 TB WD Red 5400 RPM drives (I'm not sure which ones would be "better"). The server would be running Windows 2012 R2 with data deduplication enabled. I would either use Windows Storage Spaces, RAID 6 or RAID 10.

[Shestakov]

Sounds good!
One more thing to be mentioned here: You can avoid duplication of weekly/monthly/yearly Full backups if use GFS retention of backup copy job. However GFS retention can not be implemented for the basic backup job (that`s why you need to create several jobs). If you want to save the space on the repository, you can create a backup copy job with the GFS and point it to another onsite repository.
Read more recommendations here. Thanks!