Coronavirus: Internet capable of taking the strain but service prioritisation key

The past few days have seen operators in Europe obliged to issue assurances about being able to maintain services given the almost exponential increase in home networking users, but networking and internet specialists are backing operators to maintain services – providing they adopt smart strategies.

On 20 March, BT was forced to rebut media stories suggesting that its networks would be unable to cope with the extra traffic of millions of home workers. A week later, first announcing that it was ready for the flock of users to home networks, and as tougher mitigation restrictions around Covid-19 began to take effect as more people contemplate a future of working from home for an extended period, BT felt obliged to release specific details of how it can cope with potential spikes.

BT noted that, fundamentally it had overbuilt its network to compensate for domestic high-definition streaming content, video gaming and other bandwidth-hungry applications.

It emphasised that the online conferencing services and video-calling applications that teleworkers were likely to be using consumed far less bandwidth. Even though it had seen weekday daytime traffic increase by 35-60% compared with similar days on the fixed network, peaking at 7.5Tb/s, that is still only around half the average evening peak, and nowhere near the 17.5Tb/s it had once had to handle.

Examining what it said was being called the world’s biggest work-from-home experiment, and with many wondering if the internet can handle the strain of rapid traffic growth and increased latency, David Belson, senior director of internet research and analysis at the Internet Society said that it was not likely that the extra strain on networks would cause a catastrophic failure of the Internet.

Indeed, Belson said that core internet infrastructure providers should be able to “easily” absorb the increase in traffic and demand, especially if the growth was gradual over a period of days, weeks, or months. He added that cloud infrastructure providers should also have sufficient additional compute, storage, and bandwidth capacity to enable their customers, including the e-learning, messaging, and videoconferencing tool providers, to scale their systems as necessary.

He noted that in order to keep traffic local, content delivery infrastructure from companies such as Akamai, Cloudflare, Google, Netflix and Apple is deployed in many last-mile networks. That said, on 19 March Netflix agreed to cease transmitting content in high-definition across the EU in order to reduce network strain and make sure increased teleworking was not impaired.

Looking at likely points of failure, Belson pinpointed the collaborative tools themselves rather than the networks. “The more likely place for the failures to occur is with the tools themselves, if they have not provisioned sufficient compute, storage, or bandwidth resources to deal with the increased traffic,” he said.

“Examples of such failures have already been observed in China, where servers supporting Baidu’s iQiyi streaming service, an educational application called Xuexitong, popular office applications including video conferencing applications DingTalk and WeChat Work, and online games from Tencent all reportedly crashed due to increased traffic,” said Belson. “These failures are arguably similar to those seen on e-commerce or gaming platforms on high-traffic days, where back-end systems are unable to process the higher than normal volume of orders or account activations.”

Eric Broockman, chief technology officer of Extreme Networks, added that while actions such as Netflix’s were welcome, reducing streaming rates merely alleviates the symptoms of a more fundamental problem, instead of tackling its roots -especially as a lot of other streaming, video conferencing and cloud services will continue to put unusually high volumes of data across networks. Operators also had to prioritise the types of applications running on the networks.

“[We] have noticed that some of the networks we help power have seen an increase of up to 25% in traffic over the last few weeks. In some cases that translates to an additional 1.5 TB of data that is being transmitted each second,” said Broockman.

“In an ideal world, network operators would obviously upgrade their infrastructure and invest in cloud-based solutions to make their networks as agile, resilient and flexible as possible. However, the reality is that this is a race against time for network operators that need to find a solution to this problem now, rather than in a few months or years down the line.

“In the short term, what network operators could do to reduce the pressure on their networks and ensure connectivity for all is to deprioritise non-essential traffic coming from applications such as online gaming,” he said. “This would then free up bandwidth for essential services, including voice and video traffic, and ease the pressure on the network without impacting service levels too much.”

Supporting Broockman’s argument, traffic data from communications provider Verizon released on 18 March showed that games usage had shown the greatest recent increase on its networks. The provider said that while video streaming sites such as Netflix have seen a 12% week-on-week increase, gaming had spiked by 75%. VPN usage had shot up 34%.