We have been discussing 5G and Edge architectures on this blog for quite some time. As enterprises & operators begin building edge data centers from the ground up or partnering with real estate companies or refreshing the brownfield, there are a host of activities from site identification to design to installation. This post will consider the top four intertwined system architecture considerations in deploying edge applications successfully on a hardware platform.
Where 5G meets MEC (Mobile Edge Compute) Applications
5G represents a business opportunity for operators and enterprises to refresh their data centers into hyperscale clouds. This means an ability to work with distributed computing and storage infrastructure optimized for smaller data center spaces for edge clouds.
When 5G network slicing becomes a reality, deployments will span not just one edge data center but across hundreds to serve various markets. The speed at which a net new edge data center can be installed, operationalized, and integrated with the cloud from an application standpoint will dictate how well scale can be achieved. Software designers will need to participate and collaborate with data center architects in these four key areas.
- Engineering for data volumes – This depends on the industry vertical as the diagram above shows, for instance, Gaming applications will have different needs for responsiveness, latency, and analytics as opposed to a business application. System designers have to contend with data ingestion, processing, and transportation to the Region. This will impact not just the choice of the software platform but also the design of the edge data center from a server, storage, and networking standpoint.
- Designing for both Hardware and Software Scalability – The Edge data center needs to be modular so it can help facilitate a scalable design that can be flexible. We discussed Outposts servers in a 1U/2U form factor which enables the rollout of flexible infrastructure.
- Automation across large-scale deployments – Everything about the edge cloud should be automated as much as possible. This is turning up servers, deploying software onto them, patching them efficiently, and managing networking and site interdependencies.
- A single view of the system– For day 2 operations spanning monitoring, upgrades, and new application enablement, it is key to have consistent management experiences across multi-sites. This means remote management, monitoring and easy maintenance of edge servers, networks, and applications.
5G success = Cloud-Native + Automation
What is true of 5G Core and Radio workloads is also true to MEC applications – they will run either on a hyperscaler or on-premise. And they will all be native workloads composed of microservices. As we have seen from the blogs on the DISH-AWS deployment, leading adopters will transition over from VNFs (Virtual Network Functions) to CNFs (Cloud Native Functions). My hope in this blog was to highlight Day 0 considerations before these edge data centers are built and commissioned by enterprises.
Image by Republica from Pixabay