carloscastilla - Fotolia
Software-defined storage emerged more than five years ago as storage companies took a page from software-defined networks and developed a software abstraction layer that decoupled storage control, management and policies from the underlying physical devices. Since then, SDS has become a marketing buzzword, and both storage technology and the acceptance of cloud services have made significant advances.
Today, SDS is increasingly seen as a way to create infrastructure-agnostic storage fabrics that don't lock data into its point of origin but let it more freely move to where applications are running or it can be more efficiently stored.
The Storage Networking Industry Association (SNIA) defines SDS as software-virtualized storage with a service management interface that enables multiple storage pools with the following characteristics:
- A virtualized data path that supports block, file and object interface and protocols.
- Standard management APIs to enable automated administration that reduces the time, effort, admin overhead and cost of provisioning and maintaining storage devices and services.
- Scalability to enable increasing the size, performance, availability and reliability of storage infrastructure without disrupting existing applications and users.
- Usage transparency and self-service usability that lets storage users monitor and manage their resource consumption and costs.
These characteristics were initially applied to on-premises VM infrastructure to combine storage systems, often of different types and performance profiles, and give virtualized workloads access to a unified storage pool. The advent of hybrid cloud infrastructure, in which organizations are spreading workloads across both on-premises and one or more cloud services, has prompted vendors to extend SDS technology to multi-cloud environments, often with a unified namespace for file protocols or an object pool for object storage that insulates users and applications from the deployment details.
Usage scenarios, drivers and benefits
Organizations use SDS in multi-cloud environments to prevent storage silos. Multi-cloud software-defined storage strategies provide technology that fights data gravity, facilitates the data movement among multiple cloud platforms and works with a variety of applications. Possible applications include databases (block storage); VDI (file storage); next-generation applications (object storage); backup and disaster recovery; and data analytics, machine learning and AI.
Multi-cloud software-defined storage can help organizations reduce costs particularly for long-term storage. It also can reduce Capex, rapidly scale capacity up and down to meet changing business needs, locate data closer to users or in specific countries to improve application performance and meet legal requirements, and more easily exploit innovative cloud data and AI services.
A VMware-sponsored IDC report noted that multi-cloud architectures are the "new enterprise normal," with the most common deployment (81%) using two or more public clouds (IaaS) and one or more private clouds. It's not surprising that IDC found the most frequently used cloud services are AWS, Google Cloud Platform (GCP) and Microsoft Azure. The study didn't ask specifically about storage, but most deployments likely include one or more storage services. The question for these organizations then is how to implement multi-cloud software-defined storage.
For example, Citibank supplements its more than 20 data centers with cloud services and has built a complete software-defined infrastructure platform that includes a storage fabric. While it maintains legacy Fibre Channel systems, the bank said many new services are being deployed on the new architecture, including big data, NoSQL and NewSQL data services, grid computing, virtual desktop infrastructure and private cloud services. Citibank is building out its hybrid infrastructure; it said the goal is to use its software-defined cloud infrastructure to reduce provisioning time for new applications, speed up update deployments and streamline the application development release cycle and management processes.
A range of product options
Many storage products address both general-purpose and niche multi-cloud use cases. Some examples from both traditional storage vendors and SDS specialists include:
- DataCore SANSymphony software puts a heterogeneous mix of storage devices into a shared, centrally managed storage pool and includes a rich set of enterprise data services including tiering, caching, compression, deduplication, encryption, snapshots, replication and thin provisioning. Its cloud gateway works with AWS, Azure and OpenStack services.
- Datrium Automatrix is an autonomous multi-cloud data management platform that can provide primary and secondary storage, and it stores and protects data across clouds.
- Hedvig provides support for popular storage protocols and a full set of services that work across both private infrastructure, IaaS environments and SaaS backup sites.
- IoFabric can automatically migrate data between on-premises systems and Amazon S3 and Glacier based on performance profiles, available capacity, workload demand and service-level agreements.
- MinIO is an open source, S3-compatible object store that can turn on-premises storage into an S3-like repository. It can also be extended to the public cloud -- Azure Blob, GCP, native S3 and others -- and adds a caching content delivery network to reduce traffic and improve performance.
- NetApp Ontap is an enterprise storage management system that works across on-premises hardware, AWS, Azure and GCP. Cloud Volumes Ontap supports popular protocols (block, file and NAS) and includes a full set of enterprise storage services.
- Veritas Multi-Cloud Management Platform enables IT to visualize, migrate and move data and workloads. It works across on-premises and multiple clouds, including AWS, Azure, GCP, IBM Cloud and Oracle Cloud.
Other companies with multi-cloud software-defined storage offerings include Nexenta Systems, Reduxio, Scality and StorPool Storage. Although they aren't considered SDS products per se, hybrid infrastructure stacks like Azure Stack, OpenStack and VMware vCloud also facilitate the free movement of data among different cloud environments within their respective ecosystems.
Furthermore, OpenSDS is an open source option operated by the Linux Foundation to address storage integration challenges in scale-out cloud-native environments. It aims to connect siloed data systems to build a "self-governed and intelligent data platform."
There's a variety of multi-cloud software defined storage products, but the technology is young with a constant flow of new offerings. However, there's a notable lack of consistency in the features and cloud platforms supported. Potential buyers should take care to assess their requirements against product capabilities before committing to what will be a strategic technology serving as the data backbone for their entire organization.
Software-defined storage mistakes to watch out for
Preconfigured SDS products bundled with hardware catch on