Executive Q&A: Software-Defined Storage
We spoke to Paul Speciale, chief product officer at Scality, to learn about the benefits of SDS and where the technology is headed.
- By James E. Powell
- August 21, 2020
Upside: We all know how big data and data-rich applications are driving the demand for more storage. That's traditionally meant data centers had to expand their hardware -- that is, until cloud data storage emerged as an important trend (especially after security concerns were addressed). What is software-defined storage (SDS) and how does it compare to traditional storage architectures?
Paul Speciale: Software-defined storage (SDS) has a few key principles that distinguish it from legacy-style storage arrays.
First, SDS decouples the functionality and intelligence features into software. A perfect example: data protection features used to be hardware-based functions such as those provided by a RAID controller. These are now software-based data protection features such as erasure coding and data replication.
Second, SDS allows the end user to select the hardware platform, both the vendor and the form factor. Most SDS vendors support standard x86-based server platforms (with standard IP networking connectivity) and commodity disk drives.
Ultimately, because software has no physical boundaries, SDS solutions typically enable scale-out of storage across multiple physical servers. This means a single SDS deployment can grow to very high levels of storage capacity and performance in a single logical system because the software can span across and manage tens to hundreds of servers and present them as a single system.
What benefits does SDS provide?
Separating the intelligence functions into software provides several key benefits.
First, software can be upgraded incrementally, which means that new features and capabilities that provide value to the end user can be delivered more frequently. This is a massive change from the days when new functionality was tied to new hardware development cycles. This can reduce the time between new features to weeks or months instead of years -- a true advantage. In many industries, this is an important benefit because new features for data privacy protection and cloud integration (among others) can be delivered much sooner.
Second, the end user now gains total freedom of choice to run on their preferred hardware platform. Instead of buying everything from a single vendor, the choice becomes open. Imagine the freedom to even reverse auction the hardware platform so you receive the best pricing. It also means today's choice of hardware vendor can be changed later to take advantage of the latest platform form factors, densities, and CPU performance and to ride the downward cost curves of commodity hardware components.
Third, SDS eliminates the standard vendor three-year "end of support" upgrade cycle. Buyers are no longer forced to upgrade software and hardware in unison on a vendor-driven timeline because the end user can select the best platform vendor for the storage software at any time.
Moreover, if the SDS vendor supports scale-out across an increasing number of storage servers, the end user gains additional advantages: eliminating "forklift upgrades" from old to new systems, ending data migration, and growing a system seamlessly across older and newer hardware generations (the advantage again being to ride the innovation and cost curves of commodity hardware).
Finally, SDS brings management simplicity. It can abstract (mask) the low-level details of hardware platform management from the storage administrator, such as dealing with the details of each individual disk drive or server in the system. Instead, the admin manages a single and growing pool of storage, which is a much simpler task than managing collections of low-level elements. This was a key finding in the IDC Healthcare spotlight report. Hospitals are typically constrained in IT resources, so finding ways to elevate and simplify storage management is a key part of reducing the TCO of technology and healthcare data management specifically.
Is SDS for on-premises storage only or does it work with cloud storage? If cloud providers already manage storage efficiently, what benefits does SDS offer for cloud storage (and why is it even needed)?
SDS is offered for both on-premises storage (for deployment on standard servers) and for cloud deployment on cloud server instances (virtual machines). This provides a key advantage in allowing the end user to have data storage closest to the application, whether that is a legacy-based application in the enterprise data center or a new cloud-native application running in AWS, Azure, or Google.
The other advantage of SDS in the cloud is to provide uniformity in management. Storage administrators have learned how to manage specific storage vendor products, and those skills can be maintained and leveraged in the cloud. Given the IT skills shortage, this is a significant benefit.
Can SDS keep up with today's ever-rising data volume and velocity?
Yes, some SDS offerings can grow to data volumes up to hundreds of petabytes of storage in a single system. This capability stems from the fact that software can span across multiple physical server boundaries, even across geographic boundaries. As I mentioned, many (but not all) modern SDS systems can indeed provide scale-out capabilities so that a single SDS deployment can grow across many servers.
The servers are, of course, interconnected by network components (typically just standard IP network switches and cables), but it is the SDS software that spans across these servers and provides the management of all of those servers and their disks as a single system. Storage administrators can grow the system by adding more servers as they need more capacity to address increasing data volumes. Moreover, it provides a single view of this large (and growing) storage pool and aggregates all the performance of the underlying servers to provide the processing power and throughput to address increasing data velocity.
Where is this technology headed? How will it improve in the next 6 months to 2 years?
The future of most infrastructure software (including but not limited to SDS) is to have the same software "stack" run everywhere seamlessly. This was difficult or impossible to achieve only a few years ago, but it can now be accomplished through new cloud-native software techniques and technologies, such as the use of containers and orchestration frameworks such as Kubernetes, which are now supported everywhere.
The benefit to the end user is an increasing freedom to deploy the software anywhere while keeping a single, consistent, and easy-to-manage view of data wherever it lives. This is a crucial trend and capability; we see a global proliferation of data across corporate data centers, large clouds, and the emerging need for "edge" data storage from IoT, mobile devices, and the emergence of autonomous vehicles.
Describe the SDS product or service you offer and its key distinguishing features.
Scality RING is a leading software-defined storage offering. RING offers a unique combination of distributed file and object storage capabilities that make it ideal for customers undergoing digital transformation, where legacy applications and cloud applications coexist. RING can grow from small systems to hundreds of petabytes seamlessly without disruption. RING is hardware-agnostic to support customer choice today and in the future. RING supports the highest levels of data durability to protect data for the long term. RING is distinguished in a few ways: notably it offers versatility and simplicity; it is open, future-proof, and cloud-ready; and it offers the lowest possible TCO.
[Editor's Note: As chief product officer, Paul leads Scality's global product and corporate marketing functions. Paul is responsible for defining the strategic directions for the RING scale-out file and object storage software and the Zenko multicloud data controller software solutions. He also manages the company's marketing programs.
Paul's technology industry experience ranges from software development and systems engineering to product management and marketing -- from Fortune 500 companies to venture-funded startups. In the last few years, Paul has focused on both cloud and next-gen storage technologies and has been a contributor in creating new companies focused on cloud orchestration platforms, distributed network attached storage (NAS), and object and cloud storage. In addition to working at Oracle and IBM, Paul has been an integral part of high-profile startups including Q-layer (acquired by Sun Microsystems) and Amplidata (acquired by Western Digital Corporation). You can contact Paul via Twitter.]