HyperScale Reference Architecture - Terminology
The following terms are used in HyperScale Reference Architecture.
A set of similarly configured nodes which share and store erasure coded data, constitute a block. Nodes within a block communicate over a dedicated private Storage Pool network. Nodes in a block are required to have similar hardware configuration, while nodes across blocks can be different as long as it conforms to a reference design. (See Architectural Overview for more information on the reference design.)
- Block Size
In the context of HyperScale Reference Architecture, block-size indicates the number of nodes in a block. (Should not be confused with block size in data storage, which indicates the size of a block.)
A Brick is a basic unit of storage in the file-system. It is a logical construct which can represent a single hard disk drive (HDD) or a group of HDD’s. In HyperScale storage, a single HDD is considered to be a Brick. Storage from bricks is shown as an exported directory from the server.
A trusted pool of linked computers working together, resembling a single computing resource.
- Disperse Factor
Dispersed volumes are based on erasure codes, providing space-efficient protection against disk or server failures. It stores an encoded fragment of the original file to each brick in a way that only a subset of the fragments is needed to recover the original file. The number of bricks that can be missing without losing access to data is configured by the administrator when the volume is created.
- Erasure Coding
Erasure Coding (EC) is a data protection method in which encoded, redundant data is broken into fragments (or chunks) and dispersed across storage devices. The chunks of data can reside across multiple servers, storage components and geographies. Erasure Coding provides greater hardware resilience from disk/node failures with less disk overhead than RAID methods.
- Network Bonding
It is the combining of multiple network interfaces on the same host for increased network resiliency and/or bandwidth. It is also known as Trunking or NIC Teaming. The behavior of traffic over bonded links is dependent on the bonding mode supported and used. Mode 0, Mode 1 and Mode 4 are supported for HyperScale Reference Architecture.
HyperScale nodes perform all data protection functions, similar to a traditional MediaAgent. All the nodes include partitions for index cache and Deduplicated Database (DDB). The nodes are used to create the shared storage pool which is used to store data. A standard rack server with appropriate levels of compute, memory, storage and network resources can be configured to be a "node".
Resilience (or fault-tolerance), in the context of HyperScale Reference Architecture, is the ability to continue with data protection tasks without disruptions, in the event of component failures to the platform.
The subset of disks across all nodes in a block used to house the same set of erasure coded chunks of data is a sub-volume. This is similar to a traditional storage RAID group. It is also the smallest set of HDD’s (Bricks) required to start and grow capacity in a block of nodes.
- Storage Pool
In HyperScale storage, a storage pool consists of storage from one or more blocks of nodes.
Last modified: 11/1/2019 6:36:54 PM