Results 1 to 11 of 11
  1. #1
    Lounge VIP
    Join Date
    Apr 2011
    Location
    Scotland
    Posts
    1,168
    Thanks
    44
    Thanked 134 Times in 115 Posts

    DoubleTake or ArcServe Hi Availabilty with SBS 2008

    Does anyone have experience of using DoubleTake or ArcServe in a SBS 2008 environment?

    I need to increase the robustness and availability of a small SBS 2008 system. System images are available and the server already takes dynamic backups to a NAS using BackupAssist. The NAS is onsite and can be used to recover user data relatively quickly if necessary, but it is not a disaster recovery mechanism and a recent server crash has left users feeling a little fragile due to the recovery time required.

    Fail-over is not native to SBS, but DoubleTake and ArcServe indicate they support that platform. An ideal scenario would be a fail-over box at a remote site (could this be a cloud-based VM?) to which users are switched semi-automatically if needed.

    If running either of these services to an offsite fail-over, does anyone know what bandwidth would be required? Are there any obvious pitfalls to avoid?

  2. #2
    WS Lounge VIP
    Join Date
    Dec 2009
    Location
    Earth
    Posts
    8,179
    Thanks
    47
    Thanked 983 Times in 913 Posts
    Seems an overly complex scenario for a single SBS box.
    If you have hardware mirrored disks, VSS and backup to a NAS, you should be able to have the thing up and running in no time.

    1. OS on 1 pair of disks.
    2. Exchange on a pair of disks.
    3. Data on another pair or RAID5.
    4. Snapshot C: (OS) once a week.
    5. Robocopy the data to the NAS daily.
    6. Backup Exchange to the Data volume regularly and Robocopy to the NAS.

    If the OS dies you replace the image.
    Exchange can be restored straight from the Data volume.
    If a single disk dies the server will still work until you can replace it.

    In other words, spend the money on a good bit of hardware and you don't need to spend anything on a remote site - unless you want bomb protection.

    cheers, Paul

  3. #3
    Lounge VIP
    Join Date
    Apr 2011
    Location
    Scotland
    Posts
    1,168
    Thanks
    44
    Thanked 134 Times in 115 Posts
    Thanks Paul, your comments do make sense. I had originally explained to the owners that upfront investment reduces downtime and improves reliability, but as always it's a balancing act of risk versus impact.

    The server runs RAID1 for the OS. User data is also on that array: which is not ideal and something that I could recommend be changed for the better.

    Snapshots are already run weekly with daily incrementals onto the NAS as you suggest.

    The main trouble is that the server is remote from my location and it went hard down on a weekend. Although I was aware it was offline, I had no indication of what was wrong (remote access was down too) and I had no physical access until Monday. I had to "tool-up" for a major system fail and it took some time for me to travel to site, followed by more time to bring the box back up with all services running cleanly. All in, the office was down until Monday lunchtime.

    The idea of failover is appealing in this scenario but I want to be more certain before making a proposal. I understand that these services copy changed data at the byte level, thereby significantly reducing bandwidth and that using QoS one can reduce impact still further. Also reading various reviews I've gleaned that one can shape the bandwidth in DoubleTake so that a lower proportion of available bandwidth is used during office hours. Of course, the main benefit is the near instant switch to the failover box, reducing downtime close to zero.

    I've a raised a ticket at DoubleTake for a demo/trial, but I fear a sales pitch and am interested in any insights people may have.

  4. #4
    WS Lounge VIP
    Join Date
    Dec 2009
    Location
    Earth
    Posts
    8,179
    Thanks
    47
    Thanked 983 Times in 913 Posts
    I can't see a way to speed things up because you can't really fail over without working out what is wrong and that means a visit. At best you would save a few hours at considerable expense and I would prefer to spend money making sure you have an off-site backup in case of a real disaster.

    Where would you fail over to? You need to consider that a fail-over may result in mail delay / failure as it will take a day for the MX DNS records to update for the new site and the same for a fall back. And there really is no instant fail over, someone has to decide the system has died and make the switch.

    If this is the only only outage in a year, you've made 99.995% up time.

    cheers, Paul

  5. #5
    Lounge VIP
    Join Date
    Apr 2011
    Location
    Scotland
    Posts
    1,168
    Thanks
    44
    Thanked 134 Times in 115 Posts
    I can't see a way to speed things up because you can't really fail over without working out what is wrong and that means a visit.
    Yup, but the visit could be planned and the switch be transparent to users, subject to the caveats below..

    You need to consider that a fail-over may result in mail delay / failure as it will take a day for the MX DNS records to update
    Good point, I hadn't thought of that question. The DoubleTake marketing blurb suggests that all services, settings and IP configurations are mirrored, but surely externally facing IP addresses, MX records, Reverse DNS etc can't be forged without asking for trouble. A very good question to put to the sales tea when they call.

    And there really is no instant fail over, someone has to decide the system has died and make the switch.
    My understanding (from marketing media) is that DoubleTake runs a monitor service that sends an alert allowing admin to make the switch remotely in a matter of minutes. Nonetheless, this was already a big question on my list because it's useless if I still need to travel to site to press a button in the event of a hard fail.

    If this is the only outage in a year, you've made 99.995% up time.
    I agree.......but a Monday morning with £££ going down the drain is not a time to tell the client that.



    ......Need to think about this one a bit more.

  6. #6
    New Lounger
    Join Date
    Dec 2009
    Location
    Seattle, Washington, USA
    Posts
    2
    Thanks
    0
    Thanked 1 Time in 1 Post
    You might look at Shadow Protect for SBS (http://www.storagecraft.com/shadow_protect_SBS.php). I have used it and it works extremely well.

  7. #7
    Lounge VIP
    Join Date
    Apr 2011
    Location
    Scotland
    Posts
    1,168
    Thanks
    44
    Thanked 134 Times in 115 Posts
    Hello Dellis, thanks for your suggestion.

    However, their website appears to be offline. From previous research Storage Protect appears to offer similar capabilities to BackupAssist, which is already implemented on the server in question.

    The main thing that I'm interested in is a failover capability. This is not available to SBS, but both DoubleTake and ArcServe suggest their product will deploy a mirror copy of the live server. Both suggest that the failover server can be brought online with all services and connections exactly as if they were on the live server.

  8. #8
    Super Moderator BATcher's Avatar
    Join Date
    Feb 2008
    Location
    A cultural area in SW England
    Posts
    3,414
    Thanks
    33
    Thanked 195 Times in 175 Posts
    You need Microsoft cluster server (MSCS)! That does failover at the expense of having a mirror server always available. Needs someone who knows what they are doing to implement it properly, and lots of duplicated hardware and telecoms. And £lots...

    (Whether it can be made to work with SBS is quite another matter.)
    BATcher

    Time prevents everything happening all at once...

  9. #9
    Lounge VIP
    Join Date
    Apr 2011
    Location
    Scotland
    Posts
    1,168
    Thanks
    44
    Thanked 134 Times in 115 Posts
    Thanks BATcher,

    yes, native clustering would be good, but the client doesn't want to go that route: as you say, it needs deep pockets.

    SBS does not support clustering or failover on it's own. So after my recent brush with a dead server I decided to take some heat out of the problem and research alternatives to the currently deployed BackAssist recovery. I wasn't hopeful to be honest, but I found that both DoubleTake and ArcServe suggest their products do indeed offer failover to a hot standby for SBS. Supprised?.....yup. Interested......you bet!

    What interests me is the capability of these products and whether they do indeed offer true failover; if anyone has experience with them and could shed light on delpoyment issues and pitfalls, e.g. upload bandwidth (DoubleTake sounds interesting in that area, having the ability to shape the upload network bandwidth consumption).

  10. #10
    WS Lounge VIP
    Join Date
    Dec 2009
    Location
    Earth
    Posts
    8,179
    Thanks
    47
    Thanked 983 Times in 913 Posts
    I can't see that you are any better off with these products as you will always have redundant hardware idling along for "maybe it will stop some day" protection. If you have redundant disks, dual power supplies, a UPS and a cool, clean room, your server will run for years without so much as a hiccup. It all comes down to how much you wan to spend vs how much downtime you think you can afford.

    cheers, Paul

  11. #11
    Lounge VIP
    Join Date
    Apr 2011
    Location
    Scotland
    Posts
    1,168
    Thanks
    44
    Thanked 134 Times in 115 Posts
    <snip> ......your server will run for years without so much as a hiccup
    Yes Paul, you know that and I know that, but this is not my server, so compromises have been made and I have to offer the best service I can, given the input from the owner of the box.

    As noted at the start it's a Fujitsu TX100S2, so no dual supply. UPS: yes. Dual disks: yes, running RAID1. Cool clean room: sort of....It runs in its own server cabinet in corner of a small clean white-collar office (in a cold part of the country!). Cooling is active but not dynamic. Single NIC (SBS does not support dual-NIC), managed switch and high speed dsl modem. Regular backups to NAS using BackupAssist.

    It seems the consensus here is that standard server-grade backup and recovery processes should be sufficient for SBS unless the owner wishes to invest a lot of cash.

    I guess that's the nub of the question really. The disaster recovery proposal and planning documentation noted the trade off between investment and mean-time-to-recover. Choices were made.

    Need to think walking the owner through these points, but it could be like walking a minefield with a map.....you know where not to tread, but one wrong step and kaboom!

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •