Cloud, Fog and MEC - Are you too confused by these jargons? Are the cloud service providers spinning the yarn with new offerings?We were all brain washed over the last decade that moving our data and applications to the centralized cloud data centers is the right thing to do, to save costs. Now, the same folks who bull dozed our data and applications to the centralized cloud data centers are coming back to us and telling us, why we should keep our data and applications at the edge. In this article, we'll try to understand the differences between the Cloud, the Fog and the MEC infrastructure.
The term Cloud is well understood by the industry today. It represents a collection of compute, storage and networking gear available in a centralized location such as a data center. The cloud infrastructure may be built in one or more locations to meet the scalability, reliability and performance needs of applications. Cloud infrastructure can be maintained and run by public cloud service providers such as Amazon, Microsoft, Google, IBM and Alibaba or owned by private enterprises. Hosting an application in Cloud data centers might've been apt for web or mobile app based services in the past. However, a centralized cloud infrastructure may not be the apt solution for applications such as IOT or Smart Cars - because of their ultra low latency requirements. Hence, the industry started looking at Fog and MEC solutions.
Fog doesn't replace Cloud, but it complements the existing Cloud infrastructure that is already available in many locations. Fog solves the latency and performance challenges in the cloud infrastructure. Fog architecture uses edge devices to perform significant amounts of compute, storage, and networking functions locally and routed over the Internet. Fog architecture allows the peer Fog nodes to communicate with each other and exchange data / intelligence. Fog standards are driven by OpenFog consortium. Fog infrastructure complements the several hundreds of cloud data centers that are already established. Fog connects the Cloud to "Things", by creating a hierarchical mesh in-between the Cloud and Things. Fog's objective is to improve the latencies, performance and scalability in an IOT deployment. Fog network decides to process data either locally at the edge infrastructure or in the centralized cloud infrastructure, based on the application's performance and latency requirements.
Multi-access Edge Computing (MEC) and Fog are twin brothers trying to solve the same problem - to bring the cloud infrastructure closer to the users to improve performance and latency. (Read: 10 things you should know about MEC). MEC is being standardized by ETSI group. Both Fog and MEC's standard bodies, OpenFog consortium and ETSI respectively, signed an agreement in Sep 2017 to collaborate with each other. Fog and MEC are complementary in nature. Fog architecture is hierarchical, whereas MEC architecture is flat (i.e, there may be multiple layers of Fog nodes in-between Cloud and "Things", whereas there is only one layer of MEC nodes deployed in the edge). Peer Fog nodes may communicate with each other (north-south/east-west), and may even communicate with other MEC nodes in the network. Fog connects Cloud to "Things" and specifically designed to meet the needs of IOT applications. MEC is catering to multiple industries and multiple services such as AR/VR, Live Video and Robotic surgeries, in addition to supporting IOT services.
Telcos' Cloud & MEC Play
Telecom service providers such as AT&T and Verizon, lost the public cloud services battle to players like Amazon, Google and Azure. They eventually admitted their defeat in the war and sold their cloud businesses to IBM, a few years back. Building and managing a public cloud service requires a lot of engineering and products DNA, which the telecom service providers lacked big time. The telecom service providers are used to engaging a vendor to build new infrastructure and rollout new services to customers. MEC is providing yet another opportunity for telcos to relaunch the war. With applications such as IOT, Augmented Reality (AR), Virtual Reality (VR) and Smart Cars demanding low latencies, public cloud service providers with their data centers in a few locations within the country, cannot promise sub-millisecond latencies. Service providers with Central Offices (COs) located in multiple locations within a city are better equipped to host the new-age applications that demand ultra low latencies. The battle to deploy MECs near the edge and closer to the users has just started. Cloud service providers are aggressively pitching their solutions to the telco service providers. It is not clear whether the telcos will partner/resell the services from the cloud service providers or if they would deploy their own home grown infrastructure for MEC. Who will win the game eventually? Is it going to be the cloud service provider or the telcos? We will have to wait and watch.