The IT/digital infrastructure of the DX (digital transformation) era is likely to be a hybrid of cloud computing—hosted in hyperscale data centers in metropolitan suburbs and overseas locations—and edge computing—whereby data processing happens close to or in the same location as the user.
Hyperscale data centers
Hyperscale data centers are huge data centers that use massive amounts of computing resources and functionality to process large volumes of data. They are designed with cloud service providers in mind, and many are located in the suburbs of Japan’s major cities (Tokyo, Osaka).
Edge data centers
Edge data centers use the features of edge computing to divide up distributed processing. They are located on-site or in close proximity to the user and are often small in scale. Linking multiple edge data centers into a single virtual data center may also be a viable way of increasing the resilience of regional IT and digital infrastructure.
Regional edge data centers
Regional edge data centers are edge data centers that serve a particular region. As an example, they are expected to be deployed for MEC (multi-access edge computing) purposes within telecommunications carriers’ core networks.
Local edge data centers
Local edge data centers offload processing that was previously performed on devices. They facilitate on-site processing when low latency, security, and availability are required, such as in user production control systems, stadiums, and autonomous driving.
Since the rise of the cloud, it has become commonplace to transfer data via broadband networks so that it
can be centralized and processed in data centers. Yet there also exist applications in which there is a
need for the processing to happen on or nearby the user device, because distributing the data in this
manner can increase response/processing speeds and reduce loads.
In the past, IT architecture has
repeatedly cycled through phases of centralized processing and decentralized processing. Japan is
currently seeing a wave of cloud computing (centralized), while in the US, the trend is toward edge
computing (decentralized).
Even so, edge computing, as a distributed processing architecture, does not
sit in opposition to cloud computing as a centralized processing architecture. An effective IoT and
digital environment can be formed by optimally allocating processing roles among edge devices/servers and
cloud computing.
