5G 5G and its impact on data centers
The 5G rollout holds the promise of much higher bandwidth, much lower latency, and significantly increased connection density compared with 4G. One particular result of this is the opportunity created for higher-performance IoT infrastructures. However, 5G implementation will call for some rethinking of data center design.
5G can legitimately be called a game-changing improvement over 4G; its performance enhancements are so significant that they will facilitate new applications that were not previously viable. However, 5G rollout will impact data centers and their role within wide-area communications networks, for reasons that we will discuss.
5G offers very high bandwidths, possibly exceeding 1 Gbps. Latency also becomes extremely low, achieving 10 mS or possibly down to 1 mS, compared with 20–30 mS for 4G. Another key parameter is connection density, which is planned to exceed 100,000 points/m2, compared with around 2,000 for 4G.
5G and the IoT
5G’s increased bandwidth allows it to compete with traditional broadband connections, encouraging growth in data-intensive applications like video streaming. However, 5G’s growth is also related very much to that of the Internet of Things (IoT). This is highlighted in a Gartner survey, in which 57 percent of respondents reported that their main interest in 5G was to drive IoT communications.
IoT networks are characterized by large numbers of sensor and actuator devices—often called ‘edge’ devices—distributed out in the field and embedded in the applications they are monitoring and controlling.
These applications exist within traffic control systems, manufacturing equipment, hospitals, and cars, among other environments.
The devices connect, using wireless communications to nearby hubs that process the sensor data, either to provide local control or to package data for communication over a long-haul network to a central or cloud-based facility that provides larger-scale data processing, analysis, and storage.
The advent of 5G, with its very high connection density, creates an opportunity for far more ambitious IoT schemes, with rich arrays of sensors whose data can be combined for more sophisticated insights into application status. Its low latency will also enable haptic applications like remote surgery that were previously impossible due to the delay between a surgeon’s hand movement and the remote actuator response. It will be equally critical with the advent of driverless cars, as instant control responses are essential for timely reactions to road conditions.
Impact on data centers
Data centers that have set up for 4G will have the capacity for handling 5G data. However, they will have to change their infrastructure to cater for 5G’s frequencies. 5G uses short wavelengths, which means small cells rather than large cell towers scattered around the country. These super-high frequencies (30 GHz to 300 GHz) will only work if devices are in close proximity to antennas. Therefore, it is likely we will see multiple input and output antennas (MIMOs) and many more small cells installed around public infrastructure.
Small cells have a clear advantage over traditional cell sites as they can be located in areas that lack space for large cell towers. What’s more, they are cheaper to deploy. Data centers will, however, need to be close enough to these cells to maintain 5G’s low latency performance and meet service-level agreements.
In some cases, micro data centers might even be deployed at the base of cell towers, allowing limited data processing with even faster response times for critical applications like autonomous vehicles.
These trends will likely lead to the break-up of larger data centers into smaller, more local data centers close to these cells. The new, low latency IoT-type applications such as the above remote surgery and driverless cars examples will clearly depend on data processing, communications channels, and power infrastructures that offer ground-breaking performance.
Apart from the requirement for fast response and low latency, there’s also the question of data volumes. While 5G encourages a growth in high-density edge sensor arrays, it creates a need to process the increased volumes of data they generate. Long-haul networks would be challenged to carry these increased data volumes, so it makes more sense to process raw data in a local data center, and only transmit processed results over distance.
Although smaller than traditional facilities, these data centers will need to be just as reliable, due to the critical applications they will be supporting. They will need at least Tier 3 availability, as defined by the Uptime Institute. This equates to 99.982 percent availability, or no more than 1.6 hours’ downtime per year. Some applications with even more stringent uptime requirements will be served by Tier 4 data centers, with 99.995 percent availability, or 26.3 minutes of annual downtime.