Edge Computing - the race to the edge
Edge computing is similar to cloud computing, and is in fact an expanded version of that technology. Although edge computing itself is not a new concept, it is again gaining a great deal of attention in the context of IoT. Edge computing is crucial for using ICT to link people and things across distributed use cases, as endpoint devices generate increasing amounts of raw data and cloud networks and cost models are simply not enough to accomplish this. Edge computing is capable of providing high value by analyzing data at a more granular level close to the source of the data and feeding the results back in real time. In addition to improving real-time response close to IoT endpoint devices, edge computing will also likely develop into wide area decentralized cooperative platforms for IoT and data distribution business platforms by connecting with endpoint devices. Vendors are already competing for these platforms.
The key changes in computing that are driving the race to the edge cover content delivery, data processing needs and reductions in network traffic. Content delivery requirements have driven the migration of compute to the edge since rich media was deployed to support the burgeoning mobile and internet arenas. By expanding traditional content delivery networks which distributed and staged content, IoT solution providers are adding compute and analytics platforms to support both ingested data from end points along with managing the movement of data from edge to core. This move support the enabling of local computation to speed the time for both analysis and action. Key verticals like Oil & Gas and retail, which consume and produce vast amounts of data, are adopting the edge computing model in an effort to reduce the time to action. Finally, processing at the edge reduces the amount of data which needs to traverse the network, with the desirable outcome of reducing network bandwidth costs.
Deploying edge computing, however, inverts the recent trend toward the centralization of IT functions – firstly in a reduced number of data centers through on premise and hosted deployment and subsequently with the move to the cloud. This decentralization of compute requires changes in management and maintenance of compute resources. Some level of awareness of the location of edge compute resources is necessary, especially where they exist in non-data center environments. It’s also important to consider whether the data held at the edge ultimately is needed at the core and how that data will be transported.
In adopting the edge computing model for both branch solutions and across IoT use cases, its important to consider the balance of data held at the edge and in the core, the use of those data sets (for example, strategic analysis may be challenging is 90% of the data is held at the edge). Regardless, hardware vendors, cloud service providers and CDN vendors are all expanding their edge offers – the question is whether enterprises will be ready with the skills and architectures necessary to capitalize on the capabilities offered by the edge.
