The wireless industry is betting hard on 5G’s multitude of promises of low-latency, high-bandwidth and the ability to address millions of unique devices. AT&T's latest press release ties 5G into the cloud (data center) as a new and unique feature to change the way the world operates. But it's not really new.
Edge computing (EC) reiterates an essential model of moving data processing from devices at the perimeter of the network to the cloud. Google has conducted such operations for nearly two decades – clients on smartphones, personal computer, and Chromebook act as input devices for text, voice, images, and video while the real work is done at massive data centers. Similarly, Apple and Microsoft have been in the same model, with Microsoft moving everything Windows desktop to its Azure cloud. Outside of the traditional data center space, Tesla is using its fleet of electric vehicles for daily data collection, enhancing its evolution to autonomous driving and using Wi-Fi as the broadband link.
EC is new in the sense the cellular industry is embracing the term and building a story that 5G will make existing practices better. The July 18, 2017 AT&T press release touts the ability of 5G to off-load computation into the cloud, leveraging its single-digit millisecond latency to aid emerging applications such as autonomous vehicles and augmented reality/virtual reality apps. As noted above, autonomous pioneer Tesla seems to be working perfectly well without 5G support.
Virtual reality (VR) is not and won't be a mobile app, given local computational overhead needed to run a decent experience and the need for an immersive VR environment not to put the participant into a real world wall or down a staircase. Augmented reality (AR) already works in the existing mobile technology - as evidenced by the Pokemon Go bubble. Google Glass was a consumer failure, but its recent reboot as an "Enterprise Edition" shows the true value of the technology. Numerous other AR systems operating today in vertical markets for training and maintenance applications are all operating without the need for 5G.
This isn't to say 5G won't have a substantial impact upon the data center or existing networks. Lowering latency to single digits at the edge of the network along with the potential for multi-gigabit speeds means issues further upstream will need to be addressed. If anything, 5G shifts potential bottlenecks away from the edge and wireless technology and back to the cloud. Everyone in the IT information chain will have to sharpen up network infrastructure and server response time in order to get the best out of 5G when it arrives. Talk of single digit latency means cloud services need to keep pace to avoid becoming a bottleneck.
For the SD-WAN community, 5G provides a low-latency path between the core network and data centers. Leading-edge applications such as real-time monitoring and control of robotics and drones will need the combination of 5G and SD-WAN to be truly practical.
How soon will 5G become a significant factor for data center planning? A lot of carriers are rushing into experimentation, but actual 5G standards won't be set until next year, with additions in 2019. Call it 2020 before we start to see quantities of 5G end-user gear, as opposed to some of the "pre-5G" or "5G ready" kit that will appear as carriers push the envelope for bragging rights and operational experience.
Data centers planning for a "5G ready" claim need to start sorting out network improvements now. Within the data center, the two most important areas for hardware refresh/upgrades will be in the areas of single digit latency to at least match 5G response times and at least 100Gbps network speeds to pair with 5G multi-gig requirements. Other specifications are likely to emerge as 5G is formalised, but low latency and lots of bandwidth are going to be the minimum requirements for data centers to be "5G capable."