Trends in Data Center Storage: Fusion, Standardization, and Openness in the Cloud Era

History shows that great powers inevitably merge and split; the development of data‑center storage has similarly gone through three stages of integration, separation, and re‑integration.

From IBM's first true mechanical hard drive in 1956 to the 1970s when storage and servers were tightly coupled, the industry has evolved dramatically.

With the advent of Direct‑Attached Storage (DAS), storage began to separate from servers, becoming more specialized. The rise of Storage Area Networks (SAN) in the 1980s and Network‑Attached Storage (NAS) in the 1990s reflected growing needs for centralized management and handling of unstructured, large‑capacity data.

In today's cloud era, the trend reverses again: the rise of X86 compute power and abundant network bandwidth enable Software‑Defined Storage (SDS) built on X86 servers. SDS offers openness, scalability, and cost‑effectiveness, making it well‑suited for cloud architectures and prompting a new wave of storage‑server convergence.

Nevertheless, most Chinese data centers still rely on traditional, centralized SAN and NAS solutions. While mature and stable, these systems are closed, heterogeneous, and expensive, failing to meet the diverse demands of modern users and unlikely to be fully replaced in the short term.

Emerging storage technologies are exploding, and the convergence of storage and compute is becoming mainstream. Yet legacy storage needs remain massive, creating a heterogeneous “storage soup” in data centers. We must address challenges such as standardizing and unifying storage, balancing rapid technological change with user needs, and easing the pain of transition.

Based on current conditions, three major trends in cloud‑era storage are identified:

Fusion

Storage will again merge with X86 servers – a “vertical fusion” exemplified by hyper‑converged infrastructure, which packs distributed storage, networking, and virtualization into a single standard server, simplifying data‑center architecture.

Horizontal fusion addresses storage‑to‑storage integration. IBM’s early SVC (Storage Virtualization Controller) is a classic example: it aggregates logical volumes from disparate arrays into a unified pool, then presents virtualized volumes to servers.

This approach solves the “storage silo” problem but introduces a single‑point‑of‑failure risk; therefore, dual‑active architectures are commonly employed to ensure data safety.

Standardization

Cloud data centers have shifted from device‑centric operations to service‑oriented models, focusing on business outcomes rather than hardware details. However, storage remains stuck in traditional operational modes, with heterogeneous devices from different vendors requiring separate management.

Users need a technology or standard that enables storage to be scheduled intelligently like servers and networks, eliminating fragmented management. Without such standardization, automation and intelligent storage services remain unrealistic.

Openness

Cloud computing promotes “as‑a‑service” thinking. The emerging “API economy” extends this concept, allowing not only IT resources but also product capabilities and enterprise value to be delivered via APIs.

Within data centers, this translates to “open storage”: storage should be offered as an independent service with open management APIs, allowing applications to integrate flexibly and users to define custom behaviors.

Modern storage has evolved over nearly a century, mirroring the broader IT landscape. While legacy storage still faces challenges, rapid technological progress and a more open mindset are fostering innovation and the emergence of new storage paradigms.

Recommended Reading:

What is the OpenPower Foundation and Alliance?