Flash Storage​

 View Only

Dynamic Drive Protection Failure and Recovery

By Subhasmita Sahu posted 02-17-2025 05:22

  

Introduction

Dynamic Drive Protection (DDP) provides an interleaved DDP group that can be expanded on a per-drive basis. It is possible to configure DDP groups with limited distributing data across the number of drives equal to or greater than the RAID width plus one, DDP achieves a shorter rebuild time.

Creating DDP Configuration

The available capacity of a DDP group is calculated as follows:

  •        Drive capacity × (Number of drives in a DDP group – Number of assigned spare drives)
  •        For RAID level, select either 6D+2P or 14D+2P by considering the following: 
    • o   14D+2P provides better capacity efficiency.
    • o   6D+2P offers faster rebuild.

Note: Instead of requiring spare drives, essential to rebuilding traditional RAID groups, DDP achieves spare drive-less operation by allocating spare areas within each drive.

The advantages of using DDP are as follows:

  • Flexibility in the number of drives at initial installation and expansion:
    Up to 32 drives can be set in a DDP group (equal to or greater than the RAID width plus one), optimizing the investment cost when adding a single drive.  
  • Distributed data and distributed spare function (spare drive-less):
    By distributing spare capacity across drives, the need for dedicated spare
    drives are eliminated (a spare area is embedded in each drive). This improves performance and distributes I/O, so that the number of write operations to individual drives is reduced, contributing to their longevity (for SSD (NVMe)). 
  • Distributed rebuild function (faster rebuild):
    By distributing data across the number of drives equal to or greater than the RAID width plus one, rebuild times are reduced.
    Creating a DDP group with more drives than the traditional RAID groups stabilizes performance in the event of drive failures.

Drive addition in the DDP group

  • One or more drives can be added in a DDP group; however, the following restrictions apply:   

    o   A maximum of 32 drives can be added in one DDP group.

    o   The 33rd drive can be added by configuring a new DDP group with at least nine drives, rather than adding one drive at a time.

  •          During DDP expansion, the internal data placement is reorganized through a process. However, instead of swapping all the data in the DDP group, only the data corresponding to the number of drives are added for DDP expansion. For example, one drive when only one drive is added is moved within each stripe, improving the speed of the data movement. The areas where data redistribution is completed gradually become usable, increasing the usable capacity in the pool.

DDP Group Capacity

  •         The user capacity of a DDP group (capacity excluding parity capacity) is estimated by using the following formula:
  • User capacity of the DDP group = Capacity per drive × (Number of drives in a DDP group – Number of assigned spare drives) 
  •          It is the capacity excluding the boundary adjustment area (an area for aligning with the capacity management unit of DDP) from the physical drive capacity. 

Note: even if the drive capacity and the number of drives are the same as those of the traditional RAID group, the capacity of the DDP group is not the same as the traditional RAID group because approximately 2% of the drive capacity is unusable. The increase in capacity when the number of drives increases from n to n+1 is adjusted by including it in the boundary adjustment area so that the increase in capacity remains constant regardless of the value of n. DDP Expansion Detail

DDP expansion, interruption of DDP expansion, and resuming DDP expansion are supported.

The following diagram shows the state transition when each of the commands is run:

DDP Expansion Procedure

When extending a DDP group with the following command the Parity Group will go into an extending mode after a few seconds, and you can view the progress with the get parity_grp command:

# raidcom extend parity_grp -parity_grp_id 1-1 -drive_location 0-10 0-11 0-12 0-13 0-14 0-15 -request_id auto

# raidcom get parity_grp -key opt

T GROUP  TOTAL_CAP(GB) V C E C_GROUP PWSV_S  RGID NUM_SPARE  NUM_REMAIN_SPARE  PROCESS_STATUS  PROGRESS(%)

P 1-1            21105 D E E -       -          0         1                 1  EXTENDING    

Rebalance

After parity group expansion is completed, the system will begin to rebalance the Dynamic Provision volumes within the pool.


Storage expansion is not complete until all DP volumes have been rebalanced.

Flexibility in DDP group configuration and number of drives

Traditional Parity Group

·       Initial installation and capacity expansion need to be done at the RAID group level, so adding more drives than the capacity is required. 

·       Capacity addition also needs to be done at the RAID group level.

·       Because of the limit on the number of drives, some slots in a drive box might be unavailable.
Example: In the case of a 24-slot box with 6D+2P x 2 RAID groups + 1 spare drive, a total of 17 drives are used, leaving 7 slots unused. In the case of RAID 6 (6D+2P), these remaining slots cannot be used. 

When Dynamic Drive Protection (DDP) Parity Group is used:

  •        A DDP group can be configured with any number of drives (equal to or greater than RAID width plus one), resulting in an appropriate implementation cost.
  •        Capacity can be added with one or more drives, resulting in an appropriate cost at capacity addition.
  •        DDP groups can be configured with any number of drives, allowing efficient use of slots in a drive box.

  •         Spare drives are embedded, improving I/O performance as the number of data drives increases. This contributes to drive longevity (for SSD (NVMe)) by reducing write operations through distributed I/O.
  •         Distributed data and distributed spare functions distribute data and spare areas across drives.

  •         Data is distributed across the number of drives greater than the RAID width, and spare drive capacity is automatically allocated to each drive.

  •         These advantages enable faster rebuilds.

Comparison between using DDP and not using DDP

Reference Screenshots of Failure and RecoverySingle Drive Failure and Recovery Status

During Failure PictureA screenshot of a computer

Description automatically generated

During Recovery Picture

A screenshot of a computer

Description automatically generated

Double Drive Failure status over GUI

A screenshot of a computer

Description automatically generated 

Double Drive Recovery status over GUI

A screenshot of a computer

Description automatically generated

Summary

The Dynamic Provisioning Pool offers flexibility, high performance, and efficient space utilization. Pools can be expanded on the fly to address capacity shortages. Virtual Volumes created within the DP-Pool consume pool capacity only when data resides on them or during ongoing I/O operations. Creating a DP-Pool using CCI enhances efficiency compared to performing the same task through the Service Processor.

 

1 comment
18 views

Permalink

Comments

20 days ago

nice blog. One query --> do we have a drive failure/rebuild screenshot for DDP Pool?