Software-defined storage (SDS) can completely change how storage is viewed when operating in the cloud space. SDS could downsize systems by requiring fewer storage devices.
Benefits of SDS in data centers. Image used courtesy of ESDS
Let’s take a look at this trending technology through the lens of a recent announcement in IBM’s SDS portfolio.
IBM’s Software-defined Storage
SDS provides the ability to take raw data from conventional programs and applications that would typically be stored in an external device and holds it in a space between the input and the storage. It can read and write back any of the stored data bits without exhausting any space on the storage device.
IBM has been producing an extensive memory and storage portfolio since the 1950s with their IBM 726, a magnetic tape system that could store 2 million digits per tape. Moving into the mid-1990s, IBM began implementing software-defined storage (SDS), a new form of storage that would revolutionize the industry.
In 1990, IBM launched Spectrum Scale, an SDS infrastructure solution for providing clients with data management tools to support large application workloads. Spectrum Scale was IBM’s first attempt at offering SDS-based solutions, a successful initial step into expanding their storage portfolio. Currently, Spectrum Scale can manage unstructured data for cloud and artificial intelligence analytics while streamlining workflows to position businesses and clients for future growth.
A brief overview of IBM’s Spectrum families. Screenshot used courtesy of Chris Saul and IBM [sign-up required]
IBM utilizes SDS for storage management, data protection services, and to establish robust IT infrastructures. Recently, IBM announced plans for a new container-native SDS solution and updates to its current Elastic Storage System (ESS). IBM defines a container as an open-source technology that is fully equipped to keep an environment running smoothly through programming. Containers offer engineers the flexibility of virtual machines while delivering workloads to private and public clouds without the need for external data protection components.
Integrating Hybrid Cloud and Elastic Storage Systems (ESS)
For SDS, a hybrid cloud is an essential infrastructure; it enables SDS to bridge the gap between public and cloud services from multiple cloud vendors.
Hybrid cloud architecture focuses less on physical connectivity and more on supporting the portability of workloads. The benefits of having unified cloud resources involve management tools that allow for monitoring, allocating, and managing across an open environment from a central console.
An ESS (elastic storage system) provides cloud computing services with the unique feature of expanding capacity and performance for any system requirement. Utilizing IBM Spectrum Scale, the revamped ESS 5000, and the new ESS 3200 models speed up artificial intelligence (AI) and deep learning workloads. The ESS 5000 claims to deliver 10% greater storage capacity than its predecessor, the ESS 3200, which offered 367TB of storage capacity per 2U node.
IBM’s ESS 3200. Image used courtesy of IBM
IBM’s newest addition to the storage portfolio will be the Spectrum Fusion, a container-native infrastructure that will integrate, compute, store, and network raw unstructured data. Spectrum Fusion removes the need to create duplicate data when moving data across large ecosystems. IBM plans to roll out the extended SDS version of Spectrum Fusion by mid-2022.
Denis Kennelly, General Manager, IBM Storage Systems, mentions how building, deploying, and managing applications requires fast availability of data, “from the edge to the data center to the cloud.”
To create this availability of data, there are challenges to consider, especially concerning software-defined software.
Challenges to Implementing Software-defined Software
SDS is a toolset that helps engineers and data scientists address the challenges of storing, manipulating, protecting, and analyzing heavy-data-driven workflows. Through SDS, IBM can offer storage system insights that alleviate the need to physically monitor storage capacity and performance.
At this rate, IBM pushing towards only SDS level platforms can dramatically change the memory and storage industry by lessening the amount of physical hardware needed in typical data centers and the need for hardware engineers. There will always be a need for electrical engineers, but on the SDS side of the full power spectrum, it would be a safe bet to gain a firm grasp on software engineering to develop future products. Through IBM, developers and engineers can find many SDS-based software tools to address problems when handling large databases.
When multiple cloud spaces are combined, each cloud brings forth data management problems and connectivity and compatibility issues. So far, SDS has seemed like a viable solution, but it requires some helping hands to meet the demand for fast integration between cloud spaces. Users can find IBM’s Spectrum Connect to monitor, automate, and orchestrate block storage across exhaustive multi-cloud environments.
To address the inherent questions that follow when talking about cloud storage, IBM has developed a downloadable platform, Spectrum Protect, designed for backing up data, recovery space, and emergency disaster recovery. Designers can download any toolset to enhance their system’s cloud storage. This may take away from requiring discrete devices but will still need an engineer’s evaluation and analysis of the system.
IBM’s Competition
IBM is not the only contender in this race for a complete SDS takeover. Dell has a sophisticated platform, the Dell EMC, which similarly approaches large data-driven systems as IBM. With the ability to combine multiple private and public cloud spaces, Dell’s ECS family of unstructured data management tools. A key difference is the amount of storage capacity at each rack space: the ECS EX500 can hold up to 480TB with upgrading capabilities to reach 6.1PB.
Another developer that has been drawn to SDS-based solutions is Nutanix, a business platform that combines various cloud spaces from large data centers to single desktop digital workspaces. A critical component that Nutanix uses that differs from IBM is storage virtualization. This step allows for a pool of physical space from multiple storage devices to appear as a single central console.
How will SDS change the way EEs design? Do you foresee changes to the way we work with memory moving forward? Share your insights in the comments below.
