Server-based storage is a re-emerging class of data storage that removes cost and complexity by housing storage media inside servers rather than in dedicated and custom-engineered storage arrays. Server-based storage ranges from simple SAN replacement appliances to newer classes of devices such as hyper-converged infrastructure appliances and software-defined storage products.
Even though server-based storage has most often described systems that start small and grow large, the term can also refer to simple boot disks in servers or PCI Express (PCIe) storage accelerator cards in servers. In these scenarios, the servers may not even be a part of a larger cluster. The local storage exists simply to store data local to the host or to accelerate existing storage systems by using local storage as a massive caching mechanism.
Server-based storage is steadily eating away at the market share of traditional data storage systems. According to IDC, server-based storage accounted for nearly 30% of the overall data storage market based on factory revenue in the third quarter of 2017.
Benefits of server-based storage
Data storage needs -- particularly unstructured data storage requirements -- are growing rapidly. Traditional systems that package storage controllers, disks, interfaces and firmware as proprietary storage arrays are often expensive, complex and lack openness. To combat traditional storage issues and to help companies grow their data footprints without breaking the bank, the industry and its customers have turned to commodity servers and storage hardware. The advancement of Intel core processing units (CPUs) has fueled the rise of server-based storage. Intel chips have become fast enough that most of its x86 servers can now perform storage management services that previously had to be off-loaded to proprietary chips or add-in cards.
As these data storage requirements become more critical, server-based storage helps companies grow storage capacity more efficiently. It does that by capitalizing on the ability of software defined-storage and hyper-converged technology to reduce the expense and complexity of storage.
How does server-based storage work?
Server-based storage first became a popular option for specific applications -- virtual desktop infrastructure (VDI) was an early driver for hyper-convergence -- or as storage for departments or small offices. In some cases, smaller companies are replacing their entire storage stack with server-based storage. Larger companies often get into the server-based storage market through new projects, such as new applications that require additional storage resources.
Rather than requiring businesses to buy complex and expensive arrays, server-based storage enables them to deploy storage using commodity x86 server hardware. With such hardware, users can install virtual storage appliances or full software-defined storage products, or they can turn a server node into a combined compute-storage hyper-converged device. Regardless of the method used, the result is typically a fully functional storage system that runs using one or more commodity server nodes tied together with software.
Instead of creating complex storage communications networks with dedicated infrastructure that includes technology like Fibre Channel networks, server-based storage nodes connect via standard Ethernet networks. This allows easy scalability as organizations grow and need more storage capacity. These clusters of servers work together to provide robust storage services.
A server-based storage node uses directly attached disks -- hard drives, solid-state drives (SSDs) or even PCIe cards -- to store data locally on the server. The software layered on top of this hardware transforms the server node into a full-fledged storage array, providing storage services to the organization.
Networked server-based storage
One type of server-based storage architecture is directly attached server-based storage networked into a shared storage pool. These networked storage systems combine commodity x86-based servers and directly attached disks with storage software that runs on standard servers, providing more-open storage platforms at lower prices than traditional systems.
Networked server-based storage includes virtual storage appliances, distributed file systems, object storage and Windows Server Storage Spaces. It capitalizes on virtualization, increasing storage use by pooling underutilized direct-attached storage into a shared storage resource. Some server-based storage virtualization products pair storage and virtual machine (VM) processing on the same server hardware. This enables these products to further increase server resource use, reducing data center hardware and space requirements as well as power consumption.
Other technologies driving server-based storage
More recently, software-defined storage and hyper-convergence have become the drivers of server-based storage. Hyper-convergence is a popular server-based storage option, particularly in virtual environments. Because it's managed through the hypervisor, non-storage administrators can quickly provision storage resources on hyper-converged appliances.
Server SANs and the cloud are also being used in the data center to pool storage. The cloud enables low-cost shared storage that scales. Some server-based storage vendors combine hyper-converged storage and the public cloud to use cloud-based server storage as a storage tier on top of a system's existing hardware resources and even for VM failover.
Server SAN architecture consists of multiple storage devices directly attached to separate servers and connected using high-speed interconnect. It's being used for VDI, big data and other high-performance applications. Amazon, Facebook and Google are high-profile proponents of server SAN.