none
S2D - mirror accelerated parity efficiency RRS feed

All replies

  • Hi ente_0815,

    Thanks for your post!

    >in a two node cluster with mirror accelerated parity, this article says that 7+ disks have a efficieny of max. 40%.What is the reason behind this? 

    As it explained in the article:

    "Suppose we have six capacity drives in each of two servers, and we want to create one 100 GB volume comprised of 10 GB of mirror and 90 GB of parity. Server-local two-way mirror is 50.0% efficient, meaning the 10 GB of mirror data takes 20 GB to store on each server. Mirrored to both servers, its total footprint is 40 GB. Server-local single parity, in this case, is 5/6 = 83.3% efficient, meaning the 90 GB of parity data takes 108 GB to store on each server. Mirrored to both servers, its total footprint is 216 GB. The total footprint is thus [(10 GB / 50.0%) + (90 GB / 83.3%)] × 2 = 256 GBfor 39.1% (100/256 = 39.1%)overall capacity efficiency."

    Which part do you have confuse?

    >Why is the efficiency with f.e. 16 disks not higher than with 8 disks?

    Suppose we have 8 capacity drives in each of two servers, and we want to create one 100 GB volume comprised of 10 GB of mirror and 90 GB of parity. Server-local two-way mirror is 50.0% efficient, meaning the 10 GB of mirror data takes 20 GB to store on each server. Mirrored to both servers, its total footprint is 40 GB. Server-local single parity, in this case, is 7/8 = 87.5% efficient, meaning the 90 GB of parity data takes 102 GB to store on each server. Mirrored to both servers, its total footprint is 204 GB. The total footprint is thus [(10 GB / 50.0%) + (90 GB / 87.5%)] × 2 = 244 GB, for 40.99% (100/244 = 40.99%) overall capacity efficiency."

    Suppose we have 4 capacity drives in each of two servers, and we want to create one 100 GB volume comprised of 10 GB of mirror and 90 GB of parity. Server-local two-way mirror is 50.0% efficient, meaning the 10 GB of mirror data takes 20 GB to store on each server. Mirrored to both servers, its total footprint is 40 GB. Server-local single parity, in this case, is 3/4 = 75% efficient, meaning the 90 GB of parity data takes 120 GB to store on each server. Mirrored to both servers, its total footprint is 240 GB. The total footprint is thus [(10 GB / 50.0%) + (90 GB / 75%)] × 2 = 280 GB, for 35.7% overall capacity efficiency."

    Do 16 disk (8 disk each server) is efficiency than 8 disks (4 each server).

    Thanks for your time!

    Best Regards,

    Anne


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


    Wednesday, July 8, 2020 7:33 AM
    Moderator
  • Hello Anne,

    thank you for your reply. You said:

    "Suppose we have 8 capacity drives in each of two servers, and we want to create one 100 GB volume comprised of 10 GB of mirror and 90 GB of parity. Server-local two-way mirror is 50.0% efficient, meaning the 10 GB of mirror data takes 20 GB to store on each server. Mirrored to both servers, its total footprint is 40 GB. Server-local single parity, in this case, is 7/8 = 87.5% efficient, meaning the 90 GB of parity data takes 102 GB to store on each server. Mirrored to both servers, its total footprint is 204 GB. The total footprint is thus [(10 GB / 50.0%) + (90 GB / 87.5%)] × 2 = 244 GB, for 40.99% (100/244 = 40.99%) overall capacity efficiency.""

    I totally agree with you. But if I create a volume with 7/8 disks (87,5%) or 11/12 (91,7%) the powershell output Get-StorageTier never shows more than 40% (but it should show f.e. your 40.99%).

    Can you confirm?

    Thanks

    Wednesday, July 8, 2020 10:54 AM
  • Hi ente_0815,

    >But if I create a volume with 7/8 disks (87,5%) or 11/12 (91,7%) the powershell output Get-StorageTier never shows more than 40% (but it should show f.e. your 40.99%).

    Storage Tier is a different concept from what we are talking above.

    Storage tiering is done in real-time and is used for both the way data is written, as well as the media the data is written to. In Storage Spaces Direct, the Capacity Tier is split in two: Performance and Capacity.

    The Performance Tier is part of the Capacity Tier and it is optimized for IO performance (hot data) while the rest of the Capacity Tier is optimized for Storage Efficiency (cold data).

    https://blog.cdemi.io/caching-and-storage-tiering-in-storage-spaces-direct/

    So, StorageTier and the nested-resiliency capacity usage is two things.

    If there's anything unclear, please feel free to feedback. If the reply could be of help, please mark it as answer.

    Thanks for your time! 

    Best Regards,

    Anne


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

    Friday, July 10, 2020 3:04 AM
    Moderator