IIC Journal of Innovation 17th Edition Applying Solutions at the Digital Edge | Page 8

Key Criteria to Move Cloud Workloads to the Edge
Figure 2 is a summary of the key selection criteria to consider when deciding where to locate the processing and storage for specific workloads . The criteria will be discussed in detail in subsequent sections .
The Need for Edge Compu�ng
Can ’ t run everything in the Cloud . There are latency , network bandwidth , data gravity , safety , security , reliability , resilience and privacy challenges
Edge
By adding layers of Edge Nodes , applica�ons can be par��oned to run at the op�mal network level .
Edge
Edge
Edge
Edge
Can ’ t run everything in intelligent endpoints . There are energy , space , weight , reliability , modularity , and trustworthiness challenges .
3
Fig . 2 - Key Selection Criteria for Cloud - Edge Partitioning .
Why can ’ t we run everything in cloud data centers ?
Conventional wisdom holds that cloud data centers are the optimal location to run computational workloads and host large-scale storage . This is indeed true for many applications . However , for an increasing percentage of critical IoT workloads , hosting in cloud data centers can have significant drawbacks . In general , if you can meet all system requirements by hosting an application fully within the cloud , you should , because that architecture will optimize scalability , flexibility , and deployment cost . In practical systems , it is often not a question of “ can ” or “ can ’ t , but a question of optimization of the implementation along multiple dimensions . This section will focus on some of the reasons why hosting computational workloads and storage exclusively on the cloud may not work for many classes of critical applications .
Latency
Latency is the round-trip time it takes a network to accept some input , perform computation , and produce a valid response based upon it . For IoT applications hosted in the cloud , this latency is the sum of many components . Consider a smart transportation safety use case , where a camera monitors an intersection , detects pedestrians in the traffic lanes , and attempts to apply the antilock brakes of approaching connected vehicles to avoid vehicle-pedestrian collisions . The first
- 4 - June 2021