DXfoto.com - Fotolia
The advent of next-generation applications and new IT consumption models is driving a significant shift in how networks and data centers are architected. Particularly for processor-intensive applications and large-scale networks, organizations are seeking ways to avoid sending traffic across long network routes that go between and among clouds and customer premises. To address the issue, more enterprises and telecom providers are looking to edge computing's distributed IT architecture, in which at least some processing functions are handled at the network edge as a more efficient and cost-effective alternative to more traditional, centralized data center designs.
This approach, in which a network of micro data centers processes data locally before moving the processed information to an on-premises data center or cloud facility, is particularly appealing to specific verticals, like healthcare, financial services, manufacturing and telecommunications companies running compute-intensive applications. These mini or micro modular prefabricated facilities incorporate data center features that include processing, storage, climate control, power distribution, physical security, network connectivity, fire suppression and shock absorption within a self-contained small-form-factor cabinet. These micro modular data centers are built to withstand often volatile atmospheric conditions.
While these containerized facilities have been used in some industries for years, innovation that makes it possible to compress more processing power into a smaller form factor has opened up more compelling use cases. In addition, explosive growth in areas like IoT deployments and AI is fueling demand for more micro modular data centers. Industry estimates predict dramatic modular data center expansion over the next five years, with double-digit compound annual growth. These micro data center architectures typically are comprised of fewer than 10 servers and 100 virtual machines.
Process locally, network globally
Prefabricated micro data centers are an excellent fit for a number of use cases. For now, they may be most commonly associated with IoT networks, where vast numbers of connected objects are networked together, often in far-flung places. Having a facility local to the network edge to process, store and then forward data from these devices can decrease traffic running over long network routes and improve overall efficiency and service quality.
But localized data processing opens the door to other applications for modular data centers. Compute-intensive and latency-sensitive applications that include virtual reality and other immersive applications, remote monitoring of oil and gas locations, logistics and telemedicine can all benefit from being close to a data center facility. Having a modular data center cache data locally and then forward it to a centralized facility can be especially beneficial to applications with high-performance requirements. The modular data center, which can be located on a container ship, in a manufacturing facility or near a cell tower, can process data from the applications and remove redundancies that lead to overconsumption of bandwidth.
Having a micro modular data center at the network edge could also make it easier for an organization to meet regulatory requirements, in that processing and storing data within a defined geographic perimeter can make it easier to comply with some government requirements and internal mandates.
Telecom companies can employ modular data centers at the network edge for a variety of use cases. Carriers can use edge computing for network service optimization. Used in conjunction with innovations like 5G wireless and network functions virtualization, edge computing can also help advance new service delivery models.
Telecom companies can use micro modular data center architecture to position themselves as providers of infrastructure as a service, platform as a service and network as a service. In another scenario, carriers can supply bare-metal compute and storage capabilities that are preconfigured and attach directly to their networks. Service providers can offer industry-specific options in segments like smart cities and retail.
5G wireless networks and edge computing are likely to be closely intertwined in the future. Telecom operators building out 5G networks could deploy micro modular data centers contiguous with 5G cell towers. Wireless providers can lease space in these facilities or provide hosted services to corporate clients that want a faster path onto the carrier's network.
Security challenges emphasize upfront planning
A micro modular data center promises to provide organizations with a more efficient, cost-effective and higher-performance option in some use cases. But while the model can open up a wealth of possibilities for innovation, opting to deploy micro modular data centers requires careful planning and an awareness of potential complications.
Enterprise and telecom providers need to understand the security implications of edge computing. While processing and storage will be handled closer to its origin point in a more contained manner, with less data traversing the network, having data at the edge also may make it more vulnerable. It is essential that organizations make sure the appropriate access controls, encryption and secure network connections are in place.
Selecting the appropriate micro modular data center with the right security controls and features for a particular application is crucial. A number of vendors, including Hewlett Packard Enterprise, Dell EMC, Schneider Electric and IBM, sell prefabricated mini data centers in a range of form factors. Businesses have some real concerns about vendor lock-in, given the extensive functionality built into the cabinets. This makes understanding immediate and longer-term requirements ahead of vendor selection critical.
Organizations need to recognize that a tremendous amount of coordination needs to take place with respect to data management and edge computing. While edge computing allows more data to be processed and stored close to the source, synchronization between the edge and across the core and cloud is still necessary. Without this level of harmonization, organizations are unlikely to reap the real benefits of edge computing.