Edge is often seen as a threat to cloud computing. However, the relationship between the two is more com-plementary: edge computing can support computing tasks that cannot be properly done in the cloud. With the growing usage of IoT solutions, closer Information Technology/Operational Technology integration and the integration of Industrial Control Systems in the IT stack, as well as future 5G campus solutions for low-latency applications, the need for edge solutions is set to accelerate.
Edge computing has at least four different ‘flavors’ which we define as follows.
Edge computing ‘in the narrow sense’: There are several definitions of edge computing, usually associated with concrete use cases. In general, edge computing is a layer between a physical de-vice or the physical world and the centralized data center. This could be a private or public cloud, an on-site central data center or a hosted/housed data center. Usually, the term ‘edge computing’ means that the compute unit is placed on-site near to or on the shop floor or the data sources it handles, is a standard server/storage system or appliance, and in most cases is owned by the user-organization.
Edge cloud: Edge cloud is often synonymously used for on-site public cloud systems or appliances like AWS Outposts, Microsoft Azure Stack, Oracle Cloud at Customer, or Google Anthos. The idea is to have the compute units on-site and dedicated for low latency and high bandwidth and have the managed environment of a public cloud.
Distributed cloud: Distributed clouds provide low latency compute, with the compute units not on-site but nearby (< 100 km), allowing low latency without local infrastructure while providing the man-aged environment of a public cloud. The whole infrastructure is owned by a public cloud provider. This is also the business model of some Telcos: they can open their switching centers and trans-mission towers’ own distributed cloud services. In combination with 5G, this will be a real differenti-ator.
Mobile edge computing: Similar to the distributed cloud, mobile edge computing provides very low latency compute. The compute units are not on-premise but very near-by (<5km, ideally < 1km placed in the cell tower), allowing very low latency without local infrastructure. The management can either be done by the customer/user or by a service provider or the Telco itself. This can be a valid business model for Telcos and their service provider partners. For customers, this ensures a clearer separation between the public cloud and the edge, while connectivity to on-premise infra-structures and the public cloud is a given.
teknowlogy Group believes that there are four main reasons why edge computing can play a vital role in IT architectures:
Latency: some use cases have real-time or at least near-real-time requirements, so response times can be an issue if the processing is done in a far-off data center or if some network nodes are in-volved, such as in the public cloud or other data centers, which are hundreds of kilometers away.
Connectivity: some use cases need computing power in areas where broadband connectivity or even mobile connectivity is not available, meaning local computing is needed.
Security/privacy: in some use cases data must not be distributed due to compliance rules, while in others, users just want some data to remain within their perimeter.
Connection costs and transmitted data volumes: even if connectivity is available, it can be worthwhile analyzing data at the edge and gathering only the results in a central data center or in the cloud to save connection costs.
Edge is a huge topic for our industry in 2021. If you want to learn more about our views edge computing join us in our webinar in collaboration with Reply on March 4th, 11 am. Wolfgang Schwab, Head of Cyber Security at PAC / teknowlogy will provide an overview of the different edge computing flavors and explains why user organizations should take a closer lock on edge computing also to optimize business processes and enable new business models. He will talk about general architecture considerations, i.e., how does edge computing fit in a hybrid cloud model. He will highlight the most interesting use cases for edge computing, e.g., digital factory, Digital Government & Smart Cities, smart transport, and smart energy, and will discuss potential architectural changes in the next years. To register, please click here.