This feature first appeared in the Fall 2020 issue of Certification Magazine. Click here to get your own print or digital copy.

The growth of distributed cloud computing is one of research and consulting firm Gartner's Top 10 Strategic Technology Trends for 2020. Gartner defines a distributed cloud as the distribution of public cloud services to different physical locations, while the operation, governance, updates, and evolution of the services are the responsibility of the originating public cloud provider.

This implies a geographically dispersed network of small clouds outside the cloud provider's data center, with the provider being responsible for operating, managing, and securing cloud services across the distributed cloud network. The distributed cloud model thus enables a network of multiple clouds with processing capabilities outside the cloud provider's central data center.

Distributed cloud provides for the extension of cloud services comprising a combination of hardware and software, such as applications, tools, platforms, security, services, and infrastructure, to customers' data centers and devices at the edge. Computer resources, storage, and other components at these sites are linked to the provider's cloud data centers.

According to Gartner, the cloud provider in a distributed cloud model is typically responsible for designing, architecting, distributing, managing, and updating these hardware and software components. Gartner expects that, by 2023, major cloud providers will develop a distributed ATM-like presence to deliver a subset of their cloud services for applications for which latency is critical.

Conception and growth

The emerging distributed cloud model represents an evolution of traditional centralized cloud computing to regionalized or geographically spread-out cloud computing services located as per application needs. Why is there a need to decentralize cloud computing services?

Artificial intelligence (AI), machine learning (ML), and the Internet of Things (IoT) are changing the enterprise IT environment. Increasingly, organizations are moving some of their data processing operations, along with required infrastructure, from traditional enterprise data centers and single clouds to other on-premise data centers, multi-clouds, and the edge in order to improve performance and security.

Currently, organizations deploy most of their workloads at enterprise data centers and in the traditional cloud, with roughly 10 percent of enterprise data processing taking place outside these locations. Gartner expects this to increase to over 50 percent by 2022. This trend toward distribution of cloud solutions is growing.

With edge-based computing on the increase, enterprises need to enable data processing at or near the point where data is generated on IoT-enabled or other devices at the edge. This can be challenging for enterprises with a vast national or global network. The distributed cloud model, with its network of small clouds that enable local processing, addresses this problem.

Distributed cloud computing is still in its infancy. Most experts predict that the distributed cloud model will develop in three phases:

Phase 1: The cloud provider will adopt a like-for-like hybrid approach to deliver some services in a distributed mode.

Phase 2: The provider will extend the like-for-like model to the next level. Here, the cloud provider, in collaboration with third parties, will deliver a subset of services from its centralized cloud, enabling it to reach communities through the third-party provider.

Phase 3: Communities comprising multiple organizations will share distributed cloud substations with the purpose of using cloud services at these public cloud substations.

Though Gartner emphasizes that the cloud provider will be responsible for operation, governance, updates, and evolution of the services in the distributed cloud, it does recognize that there are constraints. In the Top 10 Strategic Technology Trends for 2020 report, Gartner points out that all edge devices are not necessarily capable of running a like-for-like service that is exactly like a centralized cloud service.

Cloud providers will also need to upgrade their capabilities if they are to be considered as distributed cloud providers. A cloud provider must be capable of at least designing, configuring, dispersing, managing, and updating distributed cloud services.

In an article for infoworld.com, David Linthicum says a large number of enterprises are not ready to handle distribution of cloud solutions, and move a substantial portion of their workloads and software to the cloud. Enterprises need to be technologically prepared to deploy and operate their systems in the distributed cloud.

According to Ankur Singla, CEO of Volterra, The distributed cloud won't happen overnight, and it really won't start taking shape until 2022 and beyond.

Benefits of a distributed cloud

Let's review some of the strongest assets for organizations planning on making a switch to a distributed cloud:

Enhanced Performance and More Responsive Transmission — The distributed cloud model makes it possible for some cloud services to be provided locally, at customers' data centers or on edge devices, resulting in less latency and server congestion, and consequently a significant improvement in performance and better responsiveness.

Furthermore, transferring huge amounts of data between some edge sites and a centralized cloud data center can be economically unviable. Distributed cloud computing obviates the need for mass data transfers. Customers will have the option of specifying performance targets pertaining to latency and output in the Service Level Agreement (SLA).

Regulatory Compliance — This model will also enable enterprises to significantly improve regulatory compliance, particularly in the area of data sovereignty. For example, EU regulations require data to remain in the country. Distributed cloud services make it possible to retain data in the country where the user has generated the data.

Security � With the distributed cloud model, data will be held and maintained across several small distributed clouds. Also, enterprises will be able to retain sensitive data within their own data center or private cloud. This offers more security than the centralized model in which all data is stored and managed at a single data center.

In case of a cyberattack on a centralized data center, the data and applications located there could be compromised. Loss of customer data or unavailability of a business-critical application can have serious consequences for a company.

On the other hand, if hackers breach one small data center, only the data and applications housed there will be at risk, minimizing the threat to all user data or an entire application, thereby reducing overall business risk.

Redundancy — The distributed cloud provides higher redundancy than the centralized cloud. Positioning components at customer data centers, remote centers, and edge sites reduces the risk of network outages and enables faster recovery from disasters and loss of service.

Centralized Control of Data and Applications — Though data processing is decentralized, the distributed cloud model enables enterprises to manage all their cloud deployments with a centralized control mechanism.

Suitable for New Technologies — Distributed cloud deployments are better suited for AI and IoT applications because these typically have low-latency requirements.

Drawbacks of a distributed cloud

Initially, heterogeneity may prove a challenge for some providers. A distributed cloud environment is heterogeneous, comprising various platforms, infrastructures, and network technologies. For cloud services to run efficiently, providers must be capable of managing compatibility and interoperability across heterogeneous clouds.

Unlike a centralized cloud network, a distributed cloud environment has numerous network access points, each of which is vulnerable to attack. Administrators will need to implement uniform security policies across the distributed cloud network in order to provide visibility. It's important to select security controls that are effective, but don't have a negative effect on network performance.

Are leading cloud providers on board?

According to an article by Gartner contributor Megan Rimol, a number of cloud providers have already begun spending on processes to move some cloud solutions closer toward users of applications that require low latency.

Gartner analysts predict that, by 2023, major cloud vendors will establish an ATM-like network of micro data centers in geographical locations where a large numbers of users will be able to access some cloud services. Occasional events, such as concerts and tournaments, will be served by pop-up cloud service points.

Leading providers, such as Amazon, Microsoft, and Google are in the early stages of introducing limited cloud-like operations near or at the edge with a view to eventually enable clients to deploy and manage an application across multiple edge sites. Some examples of this, as Volterra's Singla points out, are Amazon Web Services Outposts, Google Cloud Anthos, and Microsoft Azure Arc.

Ultimately, the distributed cloud model is expected to enable all centralized cloud services to be accessible across different edge environments. This means that companies will be able to implement, operate, and maintain a host of different applications, including those on edge systems, in a number of clouds, and in legacy enterprise data centers, some of which may be decades old, as well as the necessary infrastructure as one distributed cloud.

Distributed cloud computing is currently in an emergent stage. Analysts and industry executives expect it will be a few years before organizations become capable of deploying, managing, and securing all compute components across various environments and locations through a distributed cloud.