Configuration for a 50-PC small business environment

1. offizieller Beitrag

    Hello everyone! I am a newbie at NAS solutions, but I keep on reading and learning stuff. I need your advice. I have read for many days now several topics here, and although I learn a lot, I am also a bit confused.


    I want to setup a file server in my work that will serve about 50 Windows PCs. I have a HP Proliant server, with 3 additional bays for data HDDs (1 is the OMV 5 boot drive). A capacity of 2 TB is more than enough for our case. Several samba shares and appropriate users setup will also be required.


    I was thinking about having 2 TB HDDs in software RAID-1, and another one for backup via rsync. Then, I have read a lot of negative information here, against raid-1, and I have second thoughts. Should I avoid software RAID-1, or any RAID in general? What would be the safest configuration in this case?

    a) software RAID-1 + 1 backup disk (rsync)

    b) hardware (it is supported by the HP server) RAID-1 + 1 backup disk (rsync)

    c) 1 data disk + 2 backup (rsync) disks? e.g. the 1st backing up every 2 hours, while the 2nd one at the end of the day
    d) something else


    Do you think that OMV is an appropriate solution for such a case?


    Any help will be highly appreciated. Thank you!

    • Offizieller Beitrag
    Zitat von yyovm

    Should I avoid software RAID-1, or any RAID in general?

    The use of a Raid in a home is generally not justified. What is sought with a Raid is access to data regardless of the failure of a disk. In a home it is not a problem that a family cannot see the photos for half a day until the problem is solved. Therefore, there is little point in looking for the complications of a Raid in this environment. A good scheduled backup is more than enough.

    In your case, you want to configure a server for an office with access for 50 clients. If a hard drive fails and 50 people stop working it's a problem. Therefore, I would advise setting up a Raid on this server.

    Zitat von yyovm

    What would be the safest configuration in this case?

    I would avoid solution b). The hardware Raid will be limited to the hardware, if the controller fails you need another one to recover the Raid. With a software Raid you can always recover the data on another machine with the same software.

    I think the best in this case would be solution a)

    Zitat von yyovm

    Do you think that OMV is an appropriate solution for such a case?

    I dont see why not.


    I would make sure that the server has enough capacity to handle the accesses of 50 clients. I'm not sure, but I would say it will take more RAM than usual, please correct me if not.

    • Offizieller Beitrag

    In your case, raid 1 with a backup via rsync (I would look at rsnapshot or borg for multiple copies) is the right way to go. Raid 1 will give you more IO for your 50 users. I would have another backup as well if this is important data.

    Thank you both, it is more clear now to me.


    The server has 4GB of RAM, and I hope that this is enough. But I guess I have to test and see in real life.


    I was also thinking that sw RAID-1 is preferable, since that even if the server fails, I can unplug the HDDs and plug them in another PC and I could have access to the data even with a live linux CD.


    Regarding the 2nd backup, I think that I will make another rsync backup, through LAN, to another PC, once per week, just in case.


    Thus, regarding the sw raid-1 provided by OMV, you think it is a safe solution? Since in this forum I found several posts proposing other solutions (via ZFS, BTRFS).


    And something else: in case one HDD in the RAID-1 fails, how much time would approximately take to rebuild it with a new 2 TB HDD? I tested with 2 small 80GB disks, and I took quit some time!

    • Offizieller Beitrag

    It really depends on how much your users are using the server. I generally tell people not to use raid when they don't have many users and like to shut their server down and/or spin down their drives often. If you have a 24/7 server, software raid 1 is very reliable if you are using good equipment. Hardware raid 1 is better but shouldn't be an issue on server grade equipment. If you do have a drive fail, you aren't down. The rebuild would probably take 48 hours. While the array is still available during that time, it will be slow and using it a lot will slow down the rebuild.


    OMV isn't providing software raid 1. It is the Linux kernel providing it and it is safe if used correctly. If you are worried about and/or not sure how to recover from a drive failure. Consider rsyncing from one main drive to the two others. If the main drive fails, just swing the samba share(s) over to the backup drive. Lots of options. Just need to try to find out what works best for you and your comfort level.

    • Offizieller Beitrag

    Regarding backup, I would do an incremental backup, maybe even hourly, locally on a separate disk on the same server. This makes it possible to quickly recover accidentally deleted files from the copy. It also protects you against file encryption by malware, you can revert to a previous version of the encrypted files if necessary.

    Also, I would regularly make a copy of that incremental backup to an external location, it may be another server, as you say.

    Zitat von ryecoaaron

    It really depends on how much your users are using the server. I generally tell people not to use raid when they don't have many users and like to shut their server down and/or spin down their drives often. If you have a 24/7 server, software raid 1 is very reliable if you are using good equipment. Hardware raid 1 is better but shouldn't be an issue on server grade equipment. If you do have a drive fail, you aren't down. The rebuild would probably take 48 hours. While the array is still available during that time, it will be slow and using it a lot will slow down the rebuild.

    I thing the server will be up from Monday to Friday, and off for the weekend. We have bought 3 x 2 TB Seagate Ironwolf NAS HDDs. Combining these with the HP Proliant server hopefully provides for a quite good configuration.

    Zitat von ryecoaaron

    OMV isn't providing software raid 1. It is the Linux kernel providing it and it is safe if used correctly. If you are worried about and/or not sure how to recover from a drive failure.

    Yes, I know, I have not written it the right way :)

    I think I could handle the failure of one disk in the sw raid1. Th server will be off only for the time needed to replace the faulty HDD (not hot swap for HHDs available on our server).

    Zitat von ryecoaaron

    Consider rsyncing from one main drive to the two others. If the main drive fails, just swing the samba share(s) over to the backup drive. Lots of options. Just need to try to find out what works best for you and your comfort level.

    I am not sure I understood what you mean here. Having a single disk as the main one, and backing this up to a sw raid-1? What would be the benefit from such a configuration?

    • Offizieller Beitrag
    Zitat von chente

    Regarding backup, I would do an incremental backup, maybe even hourly, locally on a separate disk on the same server. This makes it possible to quickly recover accidentally deleted files from the copy. It also protects you against file encryption by malware, you can revert to a previous version of the encrypted files if necessary.

    I agree 100%. I have had to help people who had their singular backup encrypted because they didn't notice right away. rsnapshot and borg both allow for multiple backups while deduplicating to not take up a lot of additional disk space. Just one thing to remember is never share your backup files via samba or nfs. That just invites a virus to encrypt the backup files.

    • Offizieller Beitrag
    Zitat von ryecoaaron

    Just one thing to remember is never share your backup files via samba or nfs. That just invites a virus to encrypt the backup files.

    I did not know it. Thanks. :thumbup:

    • Offizieller Beitrag
    Zitat von ryecoaaron

    Just one thing to remember is never share your backup files via samba or nfs. That just invites a virus to encrypt the backup files.

    A question. A shared folder in read-only mode also has the same danger?

    • Offizieller Beitrag
    Zitat von chente

    A shared folder in read-only mode also has the same danger?

    As long as it isn't smb1 or lanman auth, it is probably ok.

    Zitat von ness1602

    This is what i've implemented for some 200+ people companies:

    OMV with 2 disk zfs raid1

    Duplicati on windows machine -> ssh to folder specifically for a local user.


    This works like a charm,and has compression.

    I was looking at FreeNAS requirements, that uses ZFS, and they point that a minimum of 8GB of RAM is required! Would that work with OMV + ZFS + 4GB of RAM that I have?

    Zitat von donh

    ZFS is the way I would go for your raid 1. Mirror in ZFS speak. It probably will be a little short on memory though. A lot also depends on the size and type of files you are access and saving.

    Yes, I do not know if 4GB is enough to use ZFS. The majority of the files are documents and spreadsheets. Few images and scanned documents.

    • Offizieller Beitrag
    Zitat von yyovm

    Would that work with OMV + ZFS + 4GB of RAM that I have?

    OpenZFS will use whatever memory it has available. My server ran for a while on 4GB and ZFS and I had no problems, but I didn't have 50 clients accessing the server. I can't say how this will affect performance.

    • Offizieller Beitrag

    If you don't have the ram, there won't be filesystem cache in ram. So, everything will get loaded from disk. For small files that are typically opened one at a time by each user, I don't think it will make a difference.

    • Offizieller Beitrag

    ZFS provides a lot more than just cache. I am not sure how much the ZFS plugin exposes or which version of ZFS it is using. Proxmox just bumped to zfs-2.1.1-pve3 and a new feature "draid" supposedly makes resilver quicker. Haven't tried that yet. But ZFS seems to be a lot more actively developed than mdraid. Not sure how it compares to "brtfs"?

    • Offizieller Beitrag
    Zitat von donh

    Not sure how it compares to "brtfs"?

    BTRFS is the other option. It is well integrated into OMV and has data corruption protection features just like ZFS.

    The only downside to BTRFS is the use of Raid 5, they have had trouble with this for a long time. But for a mirrored setup it should work just as well as ZFS.

    • Offizieller Beitrag
    Zitat von donh

    ZFS provides a lot more than just cache. I am not sure how much the ZFS plugin exposes or which version of ZFS it is using. Proxmox just bumped to zfs-2.1.1-pve3 and a new feature "draid" supposedly makes resilver quicker. Haven't tried that yet. But ZFS seems to be a lot more actively developed than mdraid. Not sure how it compares to "brtfs"?

    I know it provides more than cache. I just said cache would suffer if there was a lack of ram.


    The plugin doesn't determine the version of ZFS. Debian does. On OMV 5 with backports enabled, you will get 2.0.3. On OMV 6, you will get 2.0.6. If you install the proxmox kernel on OMV, you will get the same version of zfs that proxmox uses.


    Being more actively developed doesn't make it better. mdraid has been around longer and at a stable point where people aren't adding more. zfs has been around a long time but people are still adding to it. btrfs is still behind. I tell people to use what they are comfortable with.

Jetzt mitmachen!

Sie haben noch kein Benutzerkonto auf unserer Seite? Registrieren Sie sich kostenlos und nehmen Sie an unserer Community teil!

Benutzerkonto erstellen Anmelden