RAID Questions. Drive swapping and configurations

Discussion in 'Mac Basics and Help' started by Jamnabot, Aug 15, 2014.

  1. Jamnabot, Aug 15, 2014
    Last edited: Aug 15, 2014

    Jamnabot macrumors newbie

    Joined:
    Nov 10, 2005
    Location:
    UK
    #1
    Hey all

    I've been dealing with my storage solution options fairly poorly up to this point and I want to start doing things much more efficiently. My only experience with RAID though is software RAIDing internal drives in my old Mac Pro tower.

    If I get a 4 bay RAID enclosure for example, can I configure the drives to work as two different pairs of drives in a RAID 1 setup? ie. Slots 1+2 are Drive 1 RAID1 and 3+4 are Drive 2 RAID1.

    At the moment I have a variety of external drives (all about 3TB in size) and I have them in pairs as a backup solution: e.g. Drive 1A is backed up manually to Drive 1B using drive comparison software and I take Drive 1B home with me as the off-site backup. It's becoming a bit of a chore to keep this up as my number of drives expands.

    If I have an external RAID setup, can I still operate my offsite backup by taking home the B drive of each RAID1 pair? Is that not advisable to be separating them frequently? Obviously the drive wouldn't be powered up while a drive is missing unless something wrong has happened.

    What about swapping disks around? If I have a 4 bay enclosure, but I have 6 drives, can I mix and match which pairs of drives are connected?
    For example, if I fire up the enclosure with Drives 1A, 1B, 3A and 3B inserted, it'll know that drives are all correctly paired? Or does it require a commitment to slots 1+2 always being Drive 1A and 1B, while 3+4 are Drive 2A and 2B?

    Sorry for all the questions. I've tried googling the subject but haven't had much luck finding the answers I'm looking for.

    EDIT:
    Looking and reading into this more, I'm beginning to see maybe I'm going about this in the wrong way.
    My old system is actually better in that I have proper BACKUPS, not just the option of taking drives off-site.
    Worst case if I delete or brake a file, I'll still have a copy on my B drive, but with RAID 1, I've killed the copy of the file too.

    In that case, maybe I should simply be thinking about an enclosure specifically set up to do timed backups/clones or the drives instead of a RAID solution.

    More thought is required
     
  2. h9826790, Aug 15, 2014
    Last edited: Aug 15, 2014

    h9826790 macrumors 604

    h9826790

    Joined:
    Apr 3, 2014
    Location:
    Hong Kong
    #2
    RAID 1 is decided to keep your system alive, but not serve as a backup solution.
     
  3. FireWire2 macrumors 6502

    FireWire2

    Joined:
    Oct 12, 2008
    #3
  4. tuna macrumors 6502

    Joined:
    Apr 11, 2010
    #4
    RAID is a deprecated technology. It never worked great because of inherent flaws in the concept and now isn't used at all any more by sophisticated storage solutions.

    For speed just use SSDs. For redundancy/backup just use a single local external drive and then maybe pay for a reputable online storage/backup service that is using modern technologies to protect data.
     
  5. phrehdd macrumors 68040

    phrehdd

    Joined:
    Oct 25, 2008
    #5
    Okay, I'll bite - what are "sophisticated storage solutions" using instead of RAID systems?
     
  6. tuna macrumors 6502

    Joined:
    Apr 11, 2010
    #6
    Different takes on Object Storage.

    Most cloud storage companies use a "replication" technique to keep copies of each of your files on 3 or more disks in 2 or more different data centers in different geographic locations so even in the case that there is a datacenter-wide disruption or even a natural disaster in any one location your data is still intact.

    That solution has relatively high performance, high availability, high redundancy, and requires relatively low computing resources, so its popular amongst cloud storage providers who need to provide not only redundancy but also reliable availability, and who can build out this storage solution using lots of cheap off-the-rack consumer grade hardware. The only downside is the requirement for lots of hard disks because obviously if each piece of data is stored on 3 disks that means 67% of the drive space is used for redundancy.

    For people who don't need as much transfer speed or availability but still want to store lots of data with unbreakable redundancy, some companies are using "fountain code" or "rateless erasure code" to set the exact amount of redundancy that they want. This provider website has some interesting details: http://amplidata.com/amplistor/architecture/#toggle-id-3

    As an example they say that you can create an array of 20 disks and set the "safety threshold" to say 4 disks so that you can have up to 4 UREs on the same file before you lose data (RAID 6 would only allow 2 UREs). They say that with that 20/4 configuration only 30% of disk space is used for redundancy. So this is an example of a post-RAID solution that optimizes disk efficiency at the expense of transfer speed and requiring more computing power overhead to run the algorithms.

    Either way, traditional RAID 5/6 is deprecated, in RAID 5's case even for relatively small home applications. RAID 5 can only have one URE before you will lose data. This means that if any one drive fails and then you have a corrupted bit on any other drive, you will lose a file.

    So like I said, there is probably some reasons to have a local backup but if you're really serious about redundancy and want more than a single local disk of backup, you should use an online backup solution that is using these more sophisticated technologies.
     
  7. phrehdd macrumors 68040

    phrehdd

    Joined:
    Oct 25, 2008
    #7
    Thanks much for taking the time to provide some info on this topic. I never considered in general RAID of any sort to be a "real back up" but as a tool for storage. Some of the schemas you mentioned kind of sit in a RAID solution set up even if the disks are not all located in the same place due to the redundancy of info on the said disks.

    I am curious, do you know if the data sent to these multiple sites start at one site then are sent to the others or is there a centralized area where data is bifurcated or similar so that some disks are independent of each other? I ask because long ago in SCSI days, we did some things similar to what you mentioned above with a reasonable level of success. (People don't recall that the SCSI command set, and parts of it by Corel were programmable.)
     
  8. RedTomato macrumors 68040

    RedTomato

    Joined:
    Mar 4, 2005
    Location:
    .. London ..
    #8
    Thank god. I hated working with RAID - it was always a recipe for tears. The number of RAID systems that I saw set up incorrectly, or worked fine for a short time before falling over when it came to expand or change a drive ....

    Thanks for the update on modern medium scale deployments. I don't work in IT any more, but I still follow the 3 copies in two places rule: a) on laptop, b) on my time capsule c) my documents / working folders also live in Dropbox.

    Query: What's the new recommendation for small scale storage eg 10TB? That's a bit too large for a pair of drives and a bit too large for online storage too.
     

Share This Page