Disaggregated storage is a new storage option that is worth watching.
Security and data privacy are among disaggregated storage's most promising features. Enterprises have been slow to embrace the new technology, but data center managers should research potential use cases of this emerging storage technology. Organizations that have different IOPS needs for certain workloads can consider disaggregated storage to ensure their storage and compute needs are met without increasing management costs.
What is disaggregated storage?
Disaggregated storage is a type of composable infrastructure that enables organizations to access logical storage pools over a network fabric such as NVMe-oF. This enables organizations to scale compute and storage elastically and independently, which makes high availability more achievable.
Disaggregation is similar to distributed storage, but there are some differences. With distributed storage, organizations purchase IT resources on demand in the cloud, often through blockchain technology. These cloud-based systems distribute user access to data on a system of centralized storage nodes. Public cloud storage provides flexible access, consumption pricing and reduced physical infrastructure.
Top uses for disaggregated storage
Disaggregation has potential uses in environments that require scale-out storage with programmatic provisioning. Although the servers operate discretely and transparently from their peers, each node has access to all resources available via the network.
Dynamic provisioning for edge development
DevOps teams could deploy disaggregated storage to dynamically reconfigure resources without adding dedicated storage. Nonlocal storage is provisioned per workload, blending cloudlike consumption with the familiarity of DAS.
Development teams often share database access. This reduces licensing and maintenance costs for traditional apps, but can cause conflicts when teams roll out microservices.
Disaggregating storage provides app teams with flexibility to choose the best option from cloud databases. Updates and testing of microservices can occur without fear of breaking base functional code.
Disaggregated storage enables organizations to access high-performance, low-latency storage and compute. This makes it easier to improve resource utilization, since organizations might need high IOPS for some workloads but not others.
There is no need to purchase, install and maintain additional hardware or software licenses.
Hyper-converged infrastructure (HCI) integrates compute, network, storage and virtualization software in a single software-defined appliance. HCI systems abstract physical hardware via hypervisor software. To expand an HCI cluster, organizations must purchase additional nodes.
Disaggregation adds another layer of abstraction to HCI and enables users to independently scale compute and storage capacity. HPE, NetApp and VMware offer disaggregated HCI options that enable this type of agility.
How disaggregated storage relates to composable infrastructure
Composable storage is an outgrowth of HCI that allows companies to buy IT infrastructure as on-demand services.
Composable storage disaggregates hardware components as shared pools. If a workload needs additional compute, networking or storage, the assets are procured from the pool and spin dynamically into service. Upon completion of the task, the resources return to the pool. The ability to compose workload resources dynamically has implications for computational storage, which pushes data operations closer to processing.
HPE helped pioneer the composable infrastructure category with its HPE Synergy product. Other products by major vendors include the Dell EMC PowerEdge MX, Lenovo ThinkAgile CP and NetApp Astra Data Store.
What's next for disaggregated storage?
Disaggregated storage technology is evolving as organizations look to find more ways to cost-effectively scale their storage resources. Startups are also leading the way with composable and computational storage products, including Burlywood, Eideticom, Fungible, Nebulon, NGD Systems, Lightbits Labs, Liqid, Pliops, SambaNova Systems, ScaleFlux and Vast Data.