Optimizing Cloud Storage Management Services

Optimizing Cloud Storage Management Services

Gabriel Alatorre (IBM Research Almaden, USA), Sandeep Gopisetty (IBM Research Almaden, USA), Divyesh Jadav (IBM Research - Almaden, USA), Bryan Langston (IBM Research Almaden, USA), Nagapramod Mandagere (IBM Research Almaden, USA), Ramani Routray (IBM Research Almaden, USA) and Heiko Ludwig (IBM Research Almaden, USA)
DOI: 10.4018/978-1-4666-8496-6.ch014
OnDemand PDF Download:
List Price: $37.50


Storage services are an essential part of an organization's IT infrastructure services and contribute a significant part of total IT costs. For this reason, various service management techniques are applied to optimize a service's storage resource usage while still addressing requirements related to performance, high availability, or disaster recovery. While storage virtualization has been the basis for many storage service management optimizations, the relatively stable environments of enterprise IT enabled all management activity to proceed in the context of change processes on specialized storage controllers. Completely virtualized environments require frequent topological changes but also enable optimized resource usage across shared resource pools. This enables lower resource and service management costs if the right storage service management architecture is deployed. This chapter focuses on cloud service management from a storage perspective, providing a set of proven methods and services to optimize storage resource usage and the management architecture that enables them.
Chapter Preview


The IT industry is experiencing massive data growth. Data is being created by an ever-increasing number of sources through the proliferation of social media technology and applications, the near ubiquitous use of mobile devices and applications, and the increasing exploitation of large-scale data analytics. Using the traditional model of acquiring, deploying, and allocating units of storage, enterprises struggle to respond to changes in business demands, which may require reduced application deployment time, increasing use of commodity components and simplification and automation of IT operations. With the advent of cloud computing, the traditional service models that deliver storage are evolving towards services that are more modular, adaptable, responsive, and interchangeable.

In response to these pressures and challenges, software defined environments (SDEs) have emerged and are rapidly evolving to create an alignment of IT with business needs where the three pillars of IT infrastructure services (Compute, Network and Storage) are programmable for the workloads. SDEs are essentially a model of defining compute, network, and storage resources with software. As a result, the orchestration of IT resources within an SDE can automatically allocate workloads to the most suitable set of infrastructure resources, accommodate for potential heterogeneity across both workloads and resources, and continuously optimize such allocations to account for changes in workload requirements.

Gone are the days of a rigid, one-to-one mapping of a piece of hardware to a single customer or workload. Today, a one-to-many, multi-tenant relationship exists among IT components, where a single piece of hardware may be consumed by multiple, even unrelated projects or customers. The advent of software defined networking means a single network switch can be logically divided up and allocated to multiple customers. A single server can be virtualized and software defined to support multiple workloads of varying degrees of resource utilization. A single storage environment may be zoned and software defined to support the needs of multiple, independent customers. Part of the necessity for SDEs is born from the need to not just manage IT, but to optimize IT. The degree of optimization is determined by the level of granularity in the software defined components. The more comprehensive the level of abstraction is, the greater the opportunity to create architectures and orchestrations that optimize the desired condition or resource.

As these resources are abstracted, they inherently give rise to the creation of new services specializing in and tailoring to the performance and optimization of the abstracted resources. This is evident through various relatively new services and products. Nicira's software-defined networking (now a part of VMware) is a full abstraction of the networking component (VMware, 2014). SwiftStack offers a software defined storage controller to manage enterprise storage (SwiftStack, 2014). Cloud Foundry is a Platform-as-a-Service which incorporates multiple software defined technologies providing an abstraction of a software development platform (Pivotal, 2014).

Beyond being the means of efficiently managing and optimizing IT resources, SDEs are proving to be the key enabler in achieving a potentially higher value proposition: business agility. As changes in business conditions occur, the ability for an organization to sense and respond to environmental conditions becomes paramount. Rather than engage in a lengthy and likely capital-intensive process to support a new business venture or experiment, a firm’s survival may depend solely on its ability to leverage the flexibility and speed an SDE affords.

Software Defined Storage (SDS) incorporates the same concepts of abstraction and virtualization as the other two pillars of software defined environments. For example, having a clear and well-defined representation of a storage system and its storage pools, tiers, service classes, and cost drivers, data can be automatically and intelligently moved to the correct tier, based on the value of the data and highly customized, customer-specific requirements. As each of these parameters are software defined, they can be optimized based on business requirements and priorities. As business process requirements evolve, so too can an organization's storage optimization strategy.

Complete Chapter List

Search this Book: