Software-Defined Storage

Software-Defined Storage

Himanshu Sahu (University of Petroleum and Energy Studies, India) and Ninni Singh (University of Petroleum and Energy Studies, India)
DOI: 10.4018/978-1-5225-3640-6.ch013

Abstract

SDS along with SDN and software-defined compute (SDC; where in computing is virtualized and software defined) creates software-defined infrastructure (SDI). SDI is the set of three components—SDN, SDS, and SDC—making a new kind of software-defined IT infrastructure where centralization and virtualization are the main focus. SDI is proposed to have infrastructure developed over commodity hardware and software stack defined over it. SDS is exploiting the same concept of decoupling and centralization in reference to storage solutions as in SDN. The SDN works on decoupling the control plane with the data plane from a layer, three switches, or router, and makes a centralized decision point called the controller. The SDS works in a similar way by moving the decision making from the storage hardware to a centralized server. It helps in developing new and existing storage solutions over the commodity storage devices. The centralization helps to create a better dynamic solution for satisfying the customized user need. The solutions are expected to be cheaper due to the use of commodity hardware.
Chapter Preview
Top

Introduction

Storage Network Industry Association (SNIA) defines SDS as

SDS is Virtualized storage with a service management interface. SDS includes pools of storage with data service characteristics that may be applied to meet the requirements specified through the service management interface. (Carlson, 2014)

The SDS is software stack that provides the centralized management interface over the commodity hardware. It can work as a part of cloud data centre or as standalone storage solution working as an end-product.

Figure 1.

Digital data growth

From 2013 to 2020, the digital universe will grow by a factor of 10, from 4.4 ZB to 44ZB. It more than doubles every two years. (Turner, 2014)

The large data required large amount of storage hardware and also there is requirement for fast and easy storage and retrieval of the data. Existing solution has improved to cope up this with innovations in the field of cloud storage (Azodolmolky, 2013). These may be considered as patch up solutions to the existing problems but they are not ready to for future. The Internet users are increasing day by day and also the data generated per users in increasing so in future the data size will become hard to manage. Not only that the information per bit is also decreasing due to rich media contents. After the emergence of IOT when each and everything will be connected to the internet the data generated will further increase (Gubbi, 2013).

Traditional Storage System

The Traditional Storage system consists of consists of storage array, magnetic disks and tape libraries. The Tradition storage system has following features.

  • Application Dedicated: In traditional storage system we use storage device dedicated to a particular application. Every application has its own storage associated with the application only.

  • Isolated Resources: The Storage resources are isolated to one to one another. This leads to wastage of resources since the underutilized resource cannot be used by other application because storage is isolated and cannot be work as a pool.

  • Policies Unique to Vertical Solutions: Since storage resources are dedicated and isolated the policy is limited to application dedicated storage.

Problems with traditional storage systems:

  • 1.

    Cost: The traditional system is not capable of creating low cost solutions that will store the huge amount of data. The maintenance, operations and support is also needed which is also not efficient in traditional system.

  • 2.

    Choice: The Traditional storage system is lacking the interoperability and flexibility due to dedicated resource to applications and unavailability of pooling of resources. The access to shared data is not possible.

  • 3.

    Cloud Deployment: Cloud deployment is not possible by using traditional storage systems since pooling of resources in not available.

Figure 2.

Traditional storage system vs. SDS

Unlike the traditional the SDS uses a software stack above the commodity hardware. The software defined stack in a centralized controlled and API is provided to develop applications over it. The isolated resources are converted into resource pools and the policies can be implemented globally (Carlson, 2014).

After considering all these factors it can be easily deduces Traditional storage management is too complex and inefficient for new data demands that storage technology not only need improvements but also a new innovation that is capable of handling the data. SDS may prove one such solution that is based on innovation and paradigm shift.

Complete Chapter List

Search this Book:
Reset