Configuring an Erasure Coding profile

You create an Erasure Coding profile by associating a storage pool with an erasure coding scheme, such as 6+3. Then, when you configure the placement instructions for an ILM rule, you can select the Erasure Coding profile. If an object matches the rule, data and parity fragments are created and distributed to the storage locations in the storage pool according to the erasure coding scheme.

Before you begin

About this task

To create an Erasure Coding profile, you associate a storage pool containing Storage Nodes with an erasure coding scheme. This association determines the number of data and parity fragments created and where the system distributes these fragments. The storage pools used in Erasure Coding profiles must include exactly one site or three or more sites. If you want to provide site redundancy, the storage pool must have at least three sites. See "What erasure coding schemes are" for more information.
Note: You must select a storage pool that contains Storage Nodes. You cannot use Archive Nodes for erasure-coded data.

After you save the Erasure Coding profile, you can change its name, but you cannot select a different storage pool or erasure coding scheme. You also cannot delete the profile.

Steps

  1. Select ILM > Erasure Coding.
    The Erasure Coding Profiles page appears.


    Erasure Coding profiles

  2. Click Create.
    The Create EC Profile dialog box appears. By default, the Storage Pool field shows All Storage Nodes and lists any available erasure coding schemes, based on the total number of Storage Nodes and sites available in your StorageGRID Webscale system.


    Create EC Profile page

  3. Enter a name for the Erasure Coding profile.
    This name is automatically appended to the storage pool name when you configure ILM rules for erasure coding. Enter a short, but meaningful name.
  4. Select a storage pool and erasure coding scheme.
    When you select a storage pool, the list of available erasure coding schemes is updated to reflect the number of Storage Nodes and sites in that pool. The following information is listed for each available scheme:
    • Erasure Code: The name of the erasure coding scheme in the following format: data fragments + parity fragments.
    • Storage Overhead (%): The additional storage required for parity fragments relative to the object's data size. Storage Overhead = Total number of parity fragments / Total number of data fragments.
    • Storage Node Redundancy: The number of Storage Nodes that can be lost while still maintaining the ability to retrieve object data.
    • Site Redundancy: Whether the selected erasure code allows the object data to be retrieved if a site is lost.

      To support site redundancy, the selected storage pool must include multiple sites, each with enough Storage Nodes to allow any site to be lost. For example, to support site redundancy using a 6+3 erasure coding scheme, the selected storage pool must include at least three sites with at least three Storage Nodes at each site.

    Messages are displayed in these cases:
    • The storage pool you selected does not provide site redundancy. The following message is expected when the selected storage pool includes only one site. You can use this Erasure Coding profile in ILM rules to protect against node failures.


      Message if site redundancy is No

    • The storage pool you selected does not satisfy the requirements for any erasure coding scheme. For example, the following message is expected when the selected storage pool includes only two sites. If you want to use erasure coding to protect object data, you must select a storage pool with exactly one site or a storage pool with three or more sites.


      Message if no erasure coding schemes are available

  5. If more than one erasure coding scheme is listed, select the one you want to use.

    When deciding which erasure coding scheme to use, you should balance fault tolerance (achieved by having more parity segments) against the network traffic requirements for repairs (more fragments equals more network traffic). For example, when deciding between a 4+2 scheme and 6+3 scheme, select the 6+3 scheme if additional parity and fault tolerance are required. Select the 4+2 scheme if network resources are constrained to reduce network usage during node repairs.

  6. Click Save.