Key Criteria to Move Cloud Workloads to the Edge Space / Weight
Space / Weight is another key constraint preventing us from hosting advanced computation workloads in many classes of IoT devices . A 1U X86 server-class computer ( a typical processor infrastructure for many IoT applications ) has a volume of about 12 liters , while many IoT devices are 1 liter or less . A 1U server may weigh about 12kg , where many IoT devices weigh hundreds of grams . You can see , if the computational workloads require processing power similar to that standard 1U server , the weight and size supported by a typical IoT device is totally inadequate . By moving a subset of the high-performance computation from the IoT device to an edge computer , we can remove many of the size and weight constraints , and provide much higher computational performance for the service .
Environmental Considerations
IoT devices often must survive harsh environmental conditions . Extended industrial temperature ranges of -40 ° C to + 85 ° C are often encountered , especially in outdoor applications , and many types of hardware grow very expensive or are impossible to implement over this wide temperature range . This is especially true of high-power CPUs , rotating disk drives , and optical interfaces . IoT devices often must survive other environmental extremes , such as the humidity , pressure , shock , contamination , vibration , and other environmental factors , as specified in standards such as MIL-STD-810 . Designing high performance computation and storage hardware to survive these extremes is very expensive . Edge computing can be located in a somewhat more protected environment , so systems can be optimized if the environmentally sensitive electronics are moved from the IoT devices to the protected edge computers .
Modularity
Modularity is another concern for many types of IoT devices . They are often not designed for reconfiguration , expansion , update , or repair . For example , to double the memory size in a commodity IoT device like a webcam , it is more cost effective to replace the entire device than to upgrade it . Edge computers can be much more modular , expandable , and configurable , allowing different computational , storage and I / O interface modules to be included as required by the specific workloads they support , and also facilitating their expansion as system requirements evolve over time . This improves the total cost of ownership of the system , allows relatively simple edge devices to continue in service much longer , and provides an evolution path for new services .
Trustworthiness
Some of the aspects of trustworthiness we discussed in conjunction with cloud-based workloads also apply to workloads run in IoT devices . For example , IoT devices may not have the energy or processing power to run strong cryptography , compromising privacy and security . IoT devices are often exposed to physical attacks , including stealing , or destroying the device . IoT devices are
- 12 - June 2021