Traditional enterprise storage is dying a slow death. The bulky old-school storage boxes that once filled the data centers of large businesses symbolize a past era. The modern data center is undergoing a tremendous transformation created by the need for business agility and fueled by software.
Let’s go back a few decades.
The concept of a separate SAN or NAS storage box outside the server/compute was created to better utilize the storage capacity in a centralized model and provide better durability, reliability and centralization of enterprise data. And the widespread adoption (EMC Symmetrix being the first one) was fueled by commoditization of a high-speed internal network.
Over a period of time, storage got tons of features for durability, reliability and scale, and simply got too expensive and bloated. And now, some key trends threaten the very existence of storage as we know it: virtualization and cost of operations; simplicity and convergence; and the cloud.
Given this trend, what will the data center and the $30 billion storage industry look like in 2025? To answer that question, we can examine the changes already underway in the data center of 2015.
Storage is often seen as an “iceberg” comprised of primary storage, the main memory needed to run active applications, and deeper secondary storage that does not directly interact with applications. Typically, secondary storage is used to back up primary storage via replication, which involves holding a secondary copy of the data. Secondary storage also can refer to cloud storage. At Druva, we see the cloud bringing flexibility in managing workloads and enabling a convergence of backup, archival and compliance so companies don’t need to run these separately.
Cloud + Convergence. When we compare cloud storage to traditional storage, the cloud-related benefits are clear. Cloud storage allows you to access your data any time, anywhere rather than carrying a physical storage drive with you. When a company moves workloads to a public cloud such as Amazon or Microsoft, for example, it only pays for the resources it uses. More workloads moved to the cloud translate to less storage needed, and better cost efficiencies. Also, keep in mind that hyper-scale cloud providers don’t buy their storage from established storage vendors, they build their own. There also are “open compute” initiatives and projects that help companies assemble cloud-based storage systems themselves.
The idea of buying storage independent of the core architecture is dying.
Also, while a traditional storage setup requires purchasing servers and the storage network separately, the move to cloud erases this problem. Some companies are buying servers, storage and network together as a data center building block, converging everything together. This convergence will make traditional storage obsolete. The idea of buying storage independent of the core architecture is dying.
In addition to the cloud and convergence, there are forward-looking storage models that are gaining momentum among enterprises. Solid-state drives (SSDs) are less expensive, more compact and more power efficient than traditional, electromechanical disks. These drives use memory chips instead of spinning platters to store information. As there are no moving parts to the drive, SSDs are smaller, last longer and consume less energy. Also, software-definedstorage (SDS) is storage managed by a software layer. This software manages policy-based provisioning and data storage management independent of the underlying hardware.
Meanwhile, software-as-a-service (SaaS) is gaining momentum among enterprises as a means to manage backups. SaaS offers cost savings in terms of personnel, hardware and physicalstorage space. A network administrator, for example, can specify which data on the network should be backed up and how often. He could sign a service-level agreement (SLA), stating that the SaaS provider would rent storage space on a cost-per-gigabyte-stored/cost-per-data-transfer basis.
If we take the SaaS/SLA model a bit further, we see how the cloud is consumerizing storage. Let’s use the iPhone as an analogy. When we install new apps on our iPhones, these apps don’t just take up the storage space needed for the app itself; these apps consume additional space for storing “documents and data,” games, pictures, videos, etc. Storage is being consumed by apps and these apps are almost more important than the hardware itself. In the coming years, storage will be like an iPhone — a consumed application with SLAs.
Basically, you will pay for a service to work as an application, just like on the iPhone you get camera functionality to use with various photo or video-related applications. The iPhone is the new cloud, where you merge various services and SLAs into one, single system. With the consumerization of cloud storage, the business focus is less on managing the infrastructure to contain and hold data, and more on building out services that bring value by leveraging the massive volumes of stored data available.
Looking out to the data center of 2025, how will things change? For one, we see today’s IT burden of allocating space, buying hardware, apportioning application storage, integrating and managing multiple products to support data storage shifting toward an environment where these services are delivered through cloud storage platforms. Such a shift, which also suggests the convergence of primary and secondary storage, will allow businesses to focus on the strategic benefit of applications, as well as building more complex and advanced applications.
(Originally published on TechCrunch on Friday October 23, 2015)