Harnessing edge data centres to local fibre and 5G networks is key
At the recent Mobile World Congress in Barcelona, Meta was pushing the fact that the current network structures will not support the dramatic improvements in latency, symmetrical bandwidth and overall speeds required for the new generation of applications they are now developing for the so-called ‘Metaverse’.
These new interactive services need to update rapidly to replicate ‘real time’ in response to their interactions. On a head mounted mixed reality display where the graphics have to respond to where someone is focusing their eyes things will have to move in single digit milliseconds. In addition the retina grade resolutions just centimetres from the eyes will need to be significantly superior – more than 4K.
Equally the more processing that is carried out ‘on premise’, such as in a AR/VR headset, will drive the unit costs high and limit demand. Power consumption will be high and will limit battery life and playing time. Other problems will be the higher level of heat as a by-product of the increased amount of processing as well as the weight and size of the units.
These problems identified with AR and VR headsets are also common across other applications and services when looking at the dilemma of where to carry out the computing required – on premise or in a centralised hyperscale data centre.
So how will today’s centralised infrastructure of networks and computing platforms manage the delivery of these services? How do you balance the challenges of both on premise computing and centralised computing with latency and high transit costs?
Here at Proximity we are actively engaged with a number of customers who are looking at edge data centres to help solve the challenge of where to locate computing power.
We are seeing a hybrid solution become more popular as it meets the twin challenges of ‘on premise’ and ‘centralised’.
An edge data centre – connected using local fibre or 5G networks reduces latency and IP transit costs. It also allows a high level of compute processing to be carried out off premise and reduce deployment costs.