Dell ObjectScale 1.3 Administration Guide

About ObjectScale object stores

ObjectScale introduces the new logical construct that is called the object store. Object stores are a discrete storage system with an individualized lifecycle and are Kubernetes (k8s) applications deployed by ObjectScale.

One ObjectScale instance can contain multiple object stores. In an ObjectScale instance, one object store can be in NVMe. Object stores are created, updated, and deleted independently from all other object stores managed by ObjectScale. As you manage the object store through its life cycle, ObjectScale interacts with the underlying Kubernetes infrastructure as needed. ObjectScale allows Kubernetes to handle the necessary changes to the cluster resources, such as storage, CPU, and other resources.

You must associate an IAM account with an object store in order to allow users within the IAM account to manage aspects of the object store. When an IAM account is added to an object store, that account becomes a tenant within that object store. A tenant is a logical construct resulting from the binding of an IAM account to an object store.

The size of the persistent volumes (PVs) which are bound to the storage server (SS) pods or NVMe engine in an object store represent the persistent storage capacity allocated for raw user data. An object store with three 200GB SS pods provides 600GB raw disk space. ObjectScale joins the persistent volumes and pods to hosts to protect data using erasure coding. Each object store has a maximum of one SS pod per k8s worker node.

To simplify the creation of object stores of the correct size and resource profile, ObjectScale uses a workload sizer tool within the workflows for creating or modifying an object store. Administrators and end users, with the appropriate permissions, can choose the correct level of resources for the object store. ObjectScale then dynamically calculates the object store requirements as you enter the required values for the object store. Supply the initial object store capacity to allow the sizing tool to correctly calculate the capacity requirements.

To support the workload inputs provided, ObjectScale then determines:

• The number of replicas in the object store.
• PV size necessary to meet the storage needs for the life of the object store.
• Any additional capacity must cover overhead, such as metadata and data protection.

In addition to the number and size of storage server instances that are required for an object store, ObjectScale also determines the size and quantity of all components that make up the object store. ObjectScale uses performance profiles to size object store resources. This release of ObjectScale includes the large performance profile.