When it finally comes about in the next two or three years, the 5G telecom standard will bring with it radical changes to the way we use data. Many people focus mostly on the consumer-grade upsides: fiber-like speeds directly to the phone, streaming virtual reality, autonomous cars, etc.
Depending on how fast people and industries adopt these innovations, however, the new features and benefits of 5G will result in a massive new strain on datacenter infrastructure. Over the next three years, IP traffic is going to triple, achieving 3.3 zettabytes globally by 2021. Much of this growth will be driven by the growth in 5G.
What’s Going to Happen in the Short Term?
The increase in global IP traffic is not going to be felt equally by everyone. Some rural areas may not see any increase at all. Dense urban areas may see a lot more. Not only will data centers take the strain, their hosted applications will as well. At VoltDB, we predict that telco databases will experience an increase in volumes and events on the order of 10-100x in areas where 5G is deployed.
Telecom companies will most likely begin to handle this problem by increasing their investment in bare-metal infrastructure – more servers, faster memory, increased compute resources. This is commensurate with recent trends towards increased data center investment. For example, in 2017, global investment in data centers reached $18.17 billion – an increase of more than $10 billion from just a year before.
Increased Hardware Investments May Not Be Enough
A new technology is gaining popularity as 5G adoption approaches – network function virtualization (NFV). NFV means taking purpose-built appliances – routers, switches, and firewalls in a telecom context – and replacing them with applications running on general-purpose servers. In a 5G context, NFV will enable a technique known as network slicing.
Network slicing enables users to run multiple virtual network segments on the same set of oh physical hardware. Each network segment can be adjusted with unique properties, which lets operators optimize these segments to best suit a varying spectrum of devices. This capability is useful for 5G, because as we’ve mentioned, 5G isn’t just for smartphones – the network will be used for autonomous cars, VR headsets, IoT devices, and more.
Just as 5G on its own is demanding, however, NFV can be resource-intensive as well. To avoid bottlenecks, the infrastructure components of NFV must be separated the metadata components that are used to shape traffic and create policies. Simply adding additional hardware isn’t going to be enough to get virtual networks past this bottleneck – rather, it will require investment in new data storage and compute techniques.
How Will Telecoms Incorporate Edge Computing?
One of the externalities of 5G is that it will enable, in some ways, the proliferation of technology that has human life in its hands. The most commonly-cited example of this is the autonomous car. 5G is going to enable autonomous cars to quickly and reliably access computing resources in the cloud, letting them more easily recognize obstacles, black ice, pedestrians, and other hazards. As a result, however, people’s lives are going to depend on how fast autonomous cars can get responses from data centers.
By way of response, telecoms may be pressured to place compute resources as close as possible to where their requests are coming from. Instead of handling all requests from a small number of central data centers, telecoms will place clusters of servers out on the edges of their network – hence, edge computing.
Under the current edge computing model, access to data collected by devices moderated by a rules engine that creates latency. At scale, the IoT model breaks down in a web of bottlenecks and many-to-many connections that are unacceptable even before 5G. After 5G, this model will be expected to absorb an even larger amount of data – autonomous cars alone are projected to produce 5 terabytes of data for every hour of driving.
Fulfilling the Promise of 5G with VoltDB
Without scalability, NFV, or an improved edge computing framework, it will be that much more difficult to deliver the benefits of 5G as they’ve been advertised. What’s more, simply throwing more and faster hardware at the problem won’t work. In order to surmount the challenges of scale, 5G telecom operators need to change the fundamental applications that allow their hardware to run.
VoltDB is a scalable relational database that’s built for high-speed, high-throughput, low-latency applications such as 5G telecom. With its ability to function as a single/multi-event processing layer, VoltDB can dramatically simplify the complex and unwieldy computing architectures that support edge computing and NFV. That’s why our customers have already chosen VoltDB as an underlying database. For more information on VoltDB and how our innovations will enable the future of telecom, begin a free trial today.