none
Is a 2-Node+Cloud Witness S2D Cluster Less Reilable than a 3-Node+Cloud Witness S2D Cluster?

    Întrebare

  • I work for a company that can easily run all our services off one host, let alone 2 or 3, but in the past before I started working here they did purchase 3 licenses of datacenter + SA so we could without needing another license. We still would like a cluster to minimize downtime in the event of a hardware failure though so I've been looking into an S2D cluster and have and S2D lab setup.

    I keep reading vague/anecdotal claims that S2D is unreliable or doesn't work right with only 2-nodes. Is there any truth to that? Microsoft documentations states that 2-node might not fail over properly without a witness or 3rd node, is that what people mean when they say 2-node is unreliable? I wouldn't consider 2-node without a witness and it doesn't make sense why anyone would?

    Of course you can't read the minds of random people on the internet, but I'm not sure what they're talking about? Is a S2D 2-node+witness less reliable than 3-node+witness? Obviously 2-node can tolerate only one failure and not two, but we have offsite hyper-v replicas already.

    Is there something in S2D's design that it simply doesn't work reliably with 2nodes+witness?

    vineri, 1 iunie 2018 18:31

Răspunsuri

  • "Does Microsoft completely support 2-node S2D deployments the same as 3-node"

    Yes, Microsoft supports both with the same support agreements.  Where have you heard that they do not?

    "is there some other reason Microsoft does not recommend 2-node S2D deployments "

    I have never heard Microsoft not recommending 2-node deployments unless there was a valid business reason.


    tim

    • Marcat ca răspuns de dmbkm luni, 25 iunie 2018 17:46
    joi, 14 iunie 2018 11:55

Toate mesajele

  • Hi,

    Have a nice day! Thanks for your question.

    1. S2D cluster support 2 nodes. Please refer to the following article for more detailed. https://docs.microsoft.com/EN-US/windows-server/storage/storage-spaces/understand-quorum

    2. Fault tolerance and storage efficiency

    Unless you have only two servers, we recommend using three-way mirroring and/or dual parity, because they offer better fault tolerance. Specifically, they ensure that all data remains safe and continuously accessible even when two fault domains – with Storage Spaces Direct, that means two servers - are affected by simultaneous failures.

    https://docs.microsoft.com/en-us/windows-server/storage/storage-spaces/storage-spaces-fault-tolerance

    Hope this helps. If you have any question or concern, please feel free to let me know.

    Best regards,

    Michael


    Please remember to mark the replies as an answers if they help.
    If you have feedback for TechNet Subscriber Support, contact tnmff@microsoft.com

    luni, 4 iunie 2018 11:05
  • Hi,

    Just checking in to see if the information provided was helpful. Please let us know if you would like further assistance.

    Best Regards,

    Michael


    Please remember to mark the replies as an answers if they help.
    If you have feedback for TechNet Subscriber Support, contact tnmff@microsoft.com

    marți, 5 iunie 2018 14:25
  • Hi,

    Could the above reply be of help? If yes, you may mark it as answer, if not, feel free to feed back.

    Best Regards,

    Michael


    Please remember to mark the replies as an answers if they help.
    If you have feedback for TechNet Subscriber Support, contact tnmff@microsoft.com

    joi, 7 iunie 2018 11:25
  • Sorry for the delay.

    No, my question has not been answered. I understand fault tolerance. I get that less servers and less drives means less resiliency, that's just the limitations of being a small business. We can afford the downtime it would take to rebuild from backups more than we can afford to buy more servers. We also already have a third server offsite with backups ready, so it makes more sense to get that server DR ready, instead of putting more money into the same site that would all go down with a power or internet outage. 

    The question is about the stability of S2D on 2-node vs 3-node cluster? And by stability I don't mean fault tolerance for hardware. I mean performance and data integrity. I've had other vendors with other software storage convergence solutions tell me S2D is not reliable on 2 nodes and their's is. Obviously I'm not going to accept their claim at face value, so I want to get a confirmation from microsoft about 2-vs-3 node support.

    Does Microsoft completely support 2-node S2D deployments the same as 3-node, or is there some other reason Microsoft does not recommend 2-node S2D deployments (aside from less hardware fault tolerance)?

    miercuri, 13 iunie 2018 16:02
  • "Does Microsoft completely support 2-node S2D deployments the same as 3-node"

    Yes, Microsoft supports both with the same support agreements.  Where have you heard that they do not?

    "is there some other reason Microsoft does not recommend 2-node S2D deployments "

    I have never heard Microsoft not recommending 2-node deployments unless there was a valid business reason.


    tim

    • Marcat ca răspuns de dmbkm luni, 25 iunie 2018 17:46
    joi, 14 iunie 2018 11:55
  • Hi,

    Thanks for your update detailed.

    There isn't recommended configuration in MS official docs. As 2-node is the same as 3-node in configuration. The quite point is that hardware fault tolerance and storge efficiency are different. So, 2-node clusters are commonly unreliable.

    In addition, the deployment requires servers with the following demands.

    1 Minimum of servers, maximum of 16 servers.

    2 Recommended that all server be the same manufacture and model.

    Hope this helps. If you have any question and concern, please feel free to let me know.

    Best regards,

    Michael


    Please remember to mark the replies as an answers if they help.
    If you have feedback for TechNet Subscriber Support, contact tnmff@microsoft.com

    sâmbătă, 16 iunie 2018 10:21
  • "The quite point is that hardware fault tolerance and storge efficiency are different. So, 2-node clusters are commonly unreliable."

    Not sure what you are trying to say here.  Very rarely is fault tolerant hardware used in a cluster, so I don't understand your statement.  I would say that more than 99.9% of the clusters configured do not use fault tolerant hardware.  And making the statement that "2-node clusters are commonly unreliable" without any data to back it up is akin to starting rumors.  Two-node clusters work just fine.  A properly configured 2-node cluster can handle any single failure just fine, so to state that they are "commonly unreliable" is a blatant mis-statement.


    tim

    duminică, 17 iunie 2018 11:11
  • Hi,

    Just want to confirm the current situations.

    Please feel free to let us know if you need further assistance.

    Best regards,

    Michael


    Please remember to mark the replies as an answers if they help.
    If you have feedback for TechNet Subscriber Support, contact tnmff@microsoft.com

    luni, 18 iunie 2018 15:39
  • "Does Microsoft completely support 2-node S2D deployments the same as 3-node"

    Yes, Microsoft supports both with the same support agreements.  Where have you heard that they do not?

    "is there some other reason Microsoft does not recommend 2-node S2D deployments "

    I have never heard Microsoft not recommending 2-node deployments unless there was a valid business reason.


    tim

    Ok thank you.

    I heard so from competitors\vendors with other storage management software solutions, however they are always vague in their reasoning why which made me skeptical. I get that S2D is pretty expensive, and that many of it's benefits can't be realized with 2-node deployments, but my employer already owns copies of DataCenter so at this point we can either use them or not use them.

    luni, 25 iunie 2018 17:46
  • I am working on the integration of different hyper-converged solutions and can tell that indeed S2D 2-node deployment can tolerate a single failure. From another hand, I met cases when disks, purchased at the same time (the same series), failed almost simultaneously in different servers. That's the case when you should have a system that can tolerate more than on failure. For example, the 2-node deployment, based on StarWind VSAN can tolerate double failures, which is only planned for 2-node S2D after Windows Server 2019 release.

    vineri, 13 iulie 2018 08:30
  • Two node is less reliable. Use three nodes or wait for the next version of Windows http://kreelbits.blogspot.com/2018/05/the-case-against-2-node-s2d-solutions.html?m=1
    • Editat de scottkreel marți, 31 iulie 2018 05:11
    marți, 31 iulie 2018 05:07