Mirroring Servers in case of emergency?

Discussion in 'Business & Enterprise Computing' started by Tony, Aug 16, 2006.

  1. Tony

    Tony New Member

    Joined:
    Jun 26, 2001
    Messages:
    9,988
    Location:
    Sydney, NSW, Australia
    eg. I have a very business critical Dell Poweredge. It does everything for this company.

    Could I possibly buy a 2nd Poweredge of the same type and mirror everything across?

    Doesn't have to be on the network constantly. Rather just mirrors everything across and able to just swap servers out in case of an hardware failure.

    Server is out of warranty and runs scsi raid5.
     
  2. Zzapped

    Zzapped Member

    Joined:
    Jul 8, 2001
    Messages:
    12,350
    Location:
    Madora Bay
    Does this existing server have any form of backup strategy ?

    Cheers

    Z
     
  3. stalin

    stalin (Taking a Break)

    Joined:
    Jun 26, 2001
    Messages:
    4,581
    Location:
    On the move
    Assuming the disks are fine and the server dies in some other way, you can just take the disks from the dead server and place it in the functioning one, and your off and running.

    Alternatley it can be another dead spare option, and your backup process can backup to your tapes, and then you can restore to the other server.

    Realtime redundancy soloution is dependant one what those critical tasks are.
     
  4. OP
    OP
    Tony

    Tony New Member

    Joined:
    Jun 26, 2001
    Messages:
    9,988
    Location:
    Sydney, NSW, Australia
    it uses tape but it has a 200Mb scsi volume so the dat72 inside only saves the email and some critical files (and not the o/s obviously)

    I like the way stalin thinks.

    There are four scsi discs in there in their hot swap trays.

    I would be thinking one spare drive in a tray would cover HDD fail contingencies.

    A whole spare server shell (no hdd) would cover any hardware faults.

    This whole line of thought came about as this server has just fallen out of warranty and I'm looking at purchasing a extended service warranty from a 3rd party but there may be another way...

    A whole spare server shell might be cheaper than a whole years' service fee!
     
  5. thetron

    thetron Member

    Joined:
    Dec 23, 2001
    Messages:
    8,167
    Location:
    Somewhere over the Rainbo
    What about a SOHO NAS or linux based NAS box?
     
  6. OP
    OP
    Tony

    Tony New Member

    Joined:
    Jun 26, 2001
    Messages:
    9,988
    Location:
    Sydney, NSW, Australia
    I am not particularly interested in the data - I use both tape and ext. hdd to save the email and work files on the server.

    What I am interested in is the downtime with this server in case of a catastrophic h/w failure and also the huge amount of time to reinstall Windows and the email program and all the other duties this server does.

    I am tempted to buy a spare HDD and also a spare server shell... off ebay even.

    The 73Gb hdd hot tray comes to just under $500. The server would probably be in the $1,000 range off ebay?

    It is just over three years old and it's dual Xeon. Not exactly cutting edge I admit but it's quite a worker.
     
    Last edited: Aug 16, 2006
  7. dogthinker

    dogthinker Member

    Joined:
    May 5, 2006
    Messages:
    1,996
    Location:
    Sydney (Glebe)
    Basically... If you want a quick recovery time you need to be performing system state backups at least. Full backups would be better still.

    Moving the HDDs from one server to another in event of
    failure is certainly quick, but your backup strategy also needs to take into the account the possibility of a failure of the RAID controller or of multiple disks. If your data isn't backed up it's not safe. If it's backed up to a tape drive in the actuall machine, and the tapes aren't rotated regularly removed you are vulnerable to risks like fire.

    If you want 99.9% uptime you should be looking at clustering, but that's going to be expensive.
     
  8. Gecko

    Gecko Member

    Joined:
    Jul 3, 2004
    Messages:
    2,716
    Location:
    Sydney
    Take a look at rsync - it lets you keep synced copies of files on 2 servers. You could sync everything onto a second server that you just change the ip of when the first one goes down.
     
  9. security

    security Member

    Joined:
    Jan 8, 2003
    Messages:
    704
    Location:
    Top of a mountain, NSW
    On ebay without drives <$500,
    8+1 20/40 DLT surestore's and the like, go for less than $200.
    30 slot LTO1 went the other day for $1500, 3/6 terabyte to play with :D

    (lavi 2 ignore list)
     
    Last edited: Aug 17, 2006
  10. lavi

    lavi Member

    Joined:
    Dec 20, 2002
    Messages:
    4,004
    Location:
    Brisbane

    yeah right! takes 4 hours to fill a LTO1 drive and I'm speaking from my experience, 80GB of local files where the LTO is attached to and another say 30GB of email from exchange via separate network card at 1GB, both servers running 172GB x2 in raid 0 then that mirrored to another 172GB x2 in Raid0.

    the reason for the raid 0 + 1 is i can take the drives out and put them in any server if shit happens.

    your 3/6tb would take a day to backup providing there is a single drive in the library :) even with 2 drives would take too long, there are quicker ways like backing up to nearline first.

    I found out servers don't just die like a normal PC, having spare server shell won't help you if your OS goes spastic one day and screws itself up.Good to have but don't rely entirely on it

    If the server can be taken offline i suggest doing a full backup of it to another spare HDD as well as daily backup on tape.
     
  11. stalin

    stalin (Taking a Break)

    Joined:
    Jun 26, 2001
    Messages:
    4,581
    Location:
    On the move
    The important thing from all of this, which i alluded to in my first post. Is what runs on the server?

    This then allows helpful advice on how to allow them to function in a high avalibility environment.

    eg. file server, you can just use DFS to have both servers share it out.
    DB's you need some shared storage, or arrange for some scheduled cold transfers.
    etc etc...
     
  12. FaTs

    FaTs Member

    Joined:
    Aug 25, 2002
    Messages:
    1,455
    Location:
    Bris Vegas...
    You could get a newer cheapish server and cluster, setup an Active/Passive cluster. About $1000 will get your a server of similar speed to your oldish Dell (one would think).

    Also do some proper backups, tape libarys aren't that expensive. ATLEAST push to a couple of external drives, you are stuffed if you loose your controller and a couple of drives.
     
  13. Disco_Stu

    Disco_Stu Member

    Joined:
    Apr 10, 2002
    Messages:
    580
    Location:
    Ipswich, Qld
    Having a spare shell, mobo, cpu ram etc would be handy, cause u would have no driver issues with just moving the HDDs across. For some of my older servers I wouldn't mind having just a spare mobo to just rip them apart, rebuild, and then be back online quickly.

    Although, A mobo mail fail, a psu may fail, anything hardware could fail, and your solution SHOULD be right. Except if its a software failure. e.g. Windows falls over and won't boot, because of corrupted system files, really damaging virus, or the RAID fails. Just keep those in mind, because they tend to be quite more time consuming than moving stuff across.

    I'd get the extra exact replica stuff (while its still available...), but also invest in maybe external HDD to copy off system state and data etc.. just incase. and they are cheap enough that even the smallest of businesses can afford a nice big one.
     
  14. OP
    OP
    Tony

    Tony New Member

    Joined:
    Jun 26, 2001
    Messages:
    9,988
    Location:
    Sydney, NSW, Australia
    just windows, lotus notes, file server, nothing fancy

    I'll give you an idea... a 3rd party contractor wants $150 a month as service fee for extended warranty on this server.

    Now granted they are providing a gold label service... they'll take the tapes and put it in an identical server on their site if a catastrophe happens.

    The server does have USB2 so I would think that a full system state backup onto an 320Gb IDE drive would be excessively simple.
     
  15. FaTs

    FaTs Member

    Joined:
    Aug 25, 2002
    Messages:
    1,455
    Location:
    Bris Vegas...
    I'm sure it has a PCI slot you can put an add in card into.
     
  16. OP
    OP
    Tony

    Tony New Member

    Joined:
    Jun 26, 2001
    Messages:
    9,988
    Location:
    Sydney, NSW, Australia
    more like usb2 external hdd case. keep it simple!
     
  17. finnigan

    finnigan Member

    Joined:
    Aug 20, 2001
    Messages:
    237
    Location:
    Adelaide
    I think Fats is referring to spare PCI slot for a USB2 PCI card.
     
  18. OP
    OP
    Tony

    Tony New Member

    Joined:
    Jun 26, 2001
    Messages:
    9,988
    Location:
    Sydney, NSW, Australia
    nah that's fine - the server is a 2003 spec. machine with USB 2.0 already

    it's a newish server - just out of warranty
     
  19. thetron

    thetron Member

    Joined:
    Dec 23, 2001
    Messages:
    8,167
    Location:
    Somewhere over the Rainbo
    Yeah i'd go for that solution best. Just one push and she's backed up

    The USB2 external case/nas can be stored offsite or a backup drive too
     
  20. finnigan

    finnigan Member

    Joined:
    Aug 20, 2001
    Messages:
    237
    Location:
    Adelaide
    Now that is mysterious.....

    The discussion went in that direction as in previous posts the server apparently didn't have USB2 and USB1 transfers were going to be 'excessively' lengthy rather than being 'excessively' simple.

    Magic? :)
     

Share This Page