Data storage architecture poised for disruption in 2020

Disruption to the data storage architecture

Atish Gude

Atish Gude, chief strategy officer, NetApp: Edge computing will advance significantly and become increasingly disruptive in the IT landscape in 2020. Edge has always been related to internet of things technologies. What's fundamentally different now is AI-driven IoT that will need the high bandwidth and low latency you get from 5G to enable instantaneous, autonomous, intelligent decision making. You will also need edge-to-cloud capabilities, through partnerships with telcos. And you need high-performance storage to manage the data and feed it into the AI engine. Storage infrastructure for edge compute will evolve rapidly to be secure and efficient and, even more importantly, connected to all cloud and core data centers and flexible enough to accept all types of unstructured and structured data formats specifically optimized for AI.

Steve McDowell

Steve McDowell, senior analyst of storage and data center technologies, Moor Insights & Strategy: Edge computing will not disrupt storage architecture in 2020, but it will in 2021. We'll see the rollout of 5G in 2020 begin to enable new edge architectures, but there will be a long cycle of experimentation as IT architects figure out how to best integrate new edge compute capabilities into their overall information architecture. Look for lots of churn, and maybe a few startups to emerge, as the industry struggles with the optimal way to integrate edge computing into enterprise IT.

Andy Walls

Andy Walls, IBM fellow, CTO and chief architect, IBM Flash Storage: Storage for AI is going to ramp up and become a critical item in 2020. As we do real-time analytics on larger data sets from different sources, we need a good scale-out storage story with low latency. AI to date has relied on in-memory databases and a lot of preparation to use the data. Where we're moving to is computational storage. Computational storage can do some of the analysis and help to determine what data is needed, instead of having to move all the data to do your analysis. Or, the storage can do some of the preliminary data sort and only send the data you need. At a bare minimum, it could be improved software that tags the data and helps to reduce the time to prepare it.

Marc Staimer

Marc Staimer, president, Dragon Slayer Consulting: Container storage is going to become a big issue in 2020, and you're going to see quite a few companies coming out of stealth specifically in this area. Container adoption is accelerating. The number of applications in enterprises that were in production on containers in 2019 was about 15%, from 0% the year before. This year, it's expected to go even higher. You can get six times the number of containers in the same hardware as you can virtual machines, and they're license-free. They're more efficient. They use less resources. Many companies moving in that direction assume they can use the same storage that they did for hypervisors. And they can't. I expect you will see several companies specializing in container storage come out of stealth this year, because I've talked to some of them. 

Coping with big data

McDowell, Moor Insights & Strategy: The rise of unstructured data, driven by both edge and AI, will elevate object storage to a first-class citizen in every storage vendor's portfolio. Object storage will no longer be relegated as just another feature of the traditional NAS appliance, but we will see dedicated all-flash object storage emerge from nearly every storage vendor in 2020.

Randy Kerns

Randy Kerns, senior strategist and analyst, Evaluator Group: Object storage will increasingly be used for some applications as primary storage. The object storage will have either all-flash storage or tiered flash storage and support select transfer of specific object data. Scale-out file systems that can scale to thousands of nodes potentially will be chosen for IT environments as the storage for advanced analytics and artificial intelligence projects that have become mainstream, business-critical IT solutions. Primarily seen in high-performance computing, scale-out file systems, including parallel file systems, are mostly new to traditional IT.

When you think of storage in 2020, you won't just think of block, file and object anymore. You'll also think stream storage.

Sudhir Srinivasan

Sudhir Srinivasan, SVP and CTO, storage division, Dell EMC: When you think of storage in 2020, you won't just think of block, file and object anymore. You'll also think stream storage. We will see new approaches to solve streaming data challenges using technologies that weren't available several years ago -- including container-based architectures and software-defined storage. Customers in different industries want to build applications that treat data as streams instead of breaking it up into separate files or objects. Data from sensors or video cameras at the edge happens as 'unbounded' streams, and customers wish to capture and process this stream data in real-time for instant analysis and insights. Ingesting and processing stream data has unique challenges that limit traditional IT and storage systems. Since streaming workloads often change throughout the day, storage capacity and compute power must be elastic to accommodate. This requires intelligence within an application that can instantly provide auto-scaling as a feature.

Data management developments

Steven Hill

Steven Hill, senior analyst, applied infrastructure and storage technologies, 451 Research: The California Consumer Privacy Act became active on Jan. 1, 2020, introducing the first state-led GDPR-like data governance to America. Given that the heart of the IT industry is based in California, it's likely that other states will enact similar legislation, and this is only the beginning of consumer-based protections that will affect the way companies need to manage their long-term data.

Staimer, Dragon Slayer Consulting: Data management is going to accelerate significantly in 2020. Think of it like software-defined data management, where it's not tied to a storage system. Today, most data management is tied to that stored system. If you take away the most valuable part of storage -- management of the data, movement of the data, protection of the data, etc. -- and make it middleware, what you've done is convert all storage into commodity. That's why the people that are doing this love it. It's not tied to a vendor-specific storage system. And it eliminates the data movement problems that people have between storage systems. Every single vendor I've talked to in the space is becoming overwhelmed with opportunity, so I expect more to jump in this year. You will also see existing vendors retool their products to deal with unstructured data management.

Service- and consumption-based storage

Jeff Kimmel

Jeff Kimmel, Primary Storage CTO, Hewlett Packard Enterprise: A confluence of factors will drive a significant uptick in as-a-service (aaS) adoption in 2020 within customer data center, colo and edge deployments, including data management and storage. Digital transformation drives workloads to be increasingly dynamic and distributed, demanding similar characteristics from the underlying infrastructure. Intelligence-driven autonomy, scalable virtualized resources, cyber security advances and capacious network and all-flash throughput increasingly enable this transformation. On-premises aaS adoption improves agility and efficiency in concert with the primacy, gravity and control of enterprise data.

Scott Sinclair

Scott Sinclair, senior analyst, Enterprise Strategy Group: In 2020, I expect a majority of IT organizations to adopt an Opex-centric, consumption-based purchase model for at least part of their on-premises storage investments. Over recent years, we have seen a shift away from the traditional capital purchasing model. Also, IT infrastructure planning and architecture continues to be a major skill shortage among IT organizations. Businesses have less time and personnel to optimize hardware purchases. It is just easier to shift to a pay-per-use model and let the vendor manage it. With nearly all the major on-premises storage vendors offering a pay-per-use model, I expect this is the year pay-per-use is adopted by a majority of organizations.