Edge Computing is a buzzword in the industry. Admittedly, the terminology faces the same challenge as other abbreviations in the field. The definition is very vague and means a slightly different thing to the model. Research firm Gartner defines edge computing “as part of a distributed computing topology where data processing is close to the edge – where things and people produce or consume this information.” The word edge in this context literally means geographical distribution. The new solution, of course, has a purpose, but there is a simple explanation for the shift in focus from the cloud:

Rapid growth is no longer expected in the cloud. Everything that has been easily moved to the cloud has already been moved. This is why most of the new growth opportunities in the “cloud” are seen on the “edge”.

Most references to the need for edge computing are unspecified applications assumed to require cloud computing with a lower transfer delay (latency). However, the most significant hype in edge computing seems to have subsided a bit. It is likely to be due to the return of service companies, particularly telecom operators, to the ground in the 5G area. So far, there are no applications or even a leading-edge architecture in the global market for the emergence of a mass of services utilizing edge computing. It applies to the envisioned architecture in which edge computing is implemented between the terminal and the cloud. The risks of erroneous investments are too significant when there is no clear payer for the solution in sight.

Intelligent traffic and edge computing

One of the best-known visions is intelligent transport and self-propelled cars. Do they reportedly need a 5G network and edge computing just because of the small transmission delay? Unfortunately, the realization of this vision does not look promising. TOP 8 self-driving technology does not rely on the mobile network outside the car. Car sensors and computers handle the matter independently, although the driver has the ultimate responsibility. Suppose the car is moving at 50 km / h and the decision has to be made at a distance of 1 meter. In that case, every telecommunication expert understands that the transmission delay of the application e2e cannot be made small enough to take the calculation and decision out of the vehicle. Of course, the mobile network can obtain some non-important information. So is the car an edge computer or a standalone computer?

Local data center and edge computing

Correspondingly, the argument of local data centers and SaaS services against the cloud is, e.g., small transmission delay. Underlying this discussion is a simple theory that a shorter geographical distance means less transmission delay and thus a better user experience. Unfortunately, this is often not the case. The share of functioning and properly dimensioned telecommunication networks in the user’s actual e2e transmission delay is marginally tiny (only a few percent of the total latency).

After analyzing numerous network environments and application response times, the better performance of global cloud vendors is evident. International cloud services are faster than local services.

Crucial here are two of the markets I claim:

  1. There is more capacity in (virtual) servers, load balancers, storage, security solutions, and data center network infrastructure with a global cloud provider than locally. (or the technology is more modern)
  2. The development, monitoring, and optimization of software architectures and software in a global cloud vendor are of better quality than local execution. It is the most decisive proportion and accounts for 80% of all transmission delays.

Edge computing does not automatically mean a small latency

Moving an application or part of an application from Central Europe or the United States to a local Nordic edge service does not automatically improve the migration delay and user experience for the reasons described above. When implementing an architecture based on edge computing or procuring services, one should consider how the actual transfer delay of the application is set and measured as a goal? – The loose theory of the advantage of short-distance communication should not rely on.

Cloud computing and the locations of global cloud services about customers are evolving rapidly, especially for Northern European customers. Not all cloud services automatically move closer but can achieve a significantly better user experience with a bit of effort.

Hannu Rokka, Senior Advisor

5Feet Networks Oy