Home > Storage > ObjectScale and ECS > Product Documentation > Dell ObjectScale Appliance Overview and Architecture > Storage topology
Administrators can install only one ObjectScale instance per Kubernetes cluster. The Kubernetes cluster consists of nodes. An ObjectScale instance can have only one object store at a time. The object stores reside on physical nodes within the Kubernetes cluster. ObjectScale appliance is deployed with the ObjectScale software bundle package by default.
In a Kubernetes cluster, a node is a physical worker node. In ObjectScale, a node is referred to as a Storage Server (SS). An SS provides disk access in an object store. At most, one SS instance from each object store is scheduled on a Kubernetes node.
SS instances are used to store user object data, which includes any associated user object metadata, and system object metadata, such as where an object is stored on disk.
The following figure depicts the Kubernetes and object store topology trees from the preceding example. Each object store is aware of its own topology only.
The following figure shows the relationship between disks and SS instances for this example. The Objectstore_1 instance independently accesses dedicated volumes and disks for data storage on Kubernetes nodes.
The figure shows each SS instance connected to a single disk for simplicity. The key point to understand is that each SS instance in an object store is tied to a single Kubernetes node. SS instances are assigned to be scheduled on a Kubernetes node. During node failure, the SS instance and underlying persistent storage are lost. The data segments in the persistent volumes that are lost are re-created across other nodes. This functionality contrasts with a common Kubernetes behavior where instances that are lost during node failure are created elsewhere. The ephemeral SS instances are associated with the persistent storage to which they attach in ObjectScale.
Data services, which are also referred to as head services, are responsible for taking client requests, extracting the required information, and passing it to the storage engine for further processing. ObjectScale supports the S3 protocol only, and uses port 80 and 443 for S3 communication, and port 4443 and 12002 for internal service communication.
Client applications including S3 Browser and Cyberduck provide a way to quickly test, or access data stored in ObjectScale.
Writing chunks in an append-only manner means that data is added or updated by keeping the original written data in place, and creating net new chunk segments. These segments may or may not be included in the chunk container of the original object. The benefit of append-only data modification is an active/active data access model which is not hindered by file-locking issues of traditional filesystems. As objects are updated or deleted, data in chunks is no longer referenced or needed. Two Garbage Collection (GC) methods are used by ObjectScale to reclaim space from discarded full chunks, or chunks containing a mixture of deleted and non-deleted object fragments which are no longer referenced: