Online gamers are obsessed with ping – the time it takes for the round trip to and from the cloud-based gaming server. If ping is above 80 milliseconds, it means you’re playing with a handicap. The larger the ping, the more chance that, no matter how well you play, you will lose.
Why ping matters
Gaming companies worry a lot about those milliseconds — they matter because they directly affect the gaming experience and, ultimately, the brand. The slightest drop in gaming response times caused by network lag can result in gamers complaining on social media and seriously damaging the company’s reputation.
Network lag has complex causes. The simplest solution is to put the hosting servers closer to the players; thus, most gaming companies maintain servers on multiple continents. But distance aside, there are also issues to do with the architecture of IP networks and the internet that make solving the issue of ping non-trivial. With massively multiplayer online role-playing games (MMORPGs) attracting millions of online gamers, just where to put those servers to optimize ping time becomes a critical consideration for IP network architecture.
Rethinking gaming architecture
Gaming is one of several applications that are changing our thinking about network architectures. Machine-to-machine communications typical in industrial automation and autonomous transport are also extremely sensitive to the issue of network latency or lag. As an answer, the industry is promoting edge computing and edge clouds, which put the compute resources much closer to the end-user and make extremely low latencies possible (under 10 milliseconds). This is, for instance, one of the key features of 5G and why it is getting so much attention for industrial applications as well as mobile cloud gaming.
This is introducing some new players into the networking and cloud space. For video streaming and e-commerce, the central cloud players have dominated because issues of latency were unimportant, and economies of scale dictated centralization. For edge computing, co-location providers are already well placed, many of them having hundreds of facilities in key markets worldwide. They not only can bring gaming servers closer to gamers, but they can also give gaming companies direct access to peering and interconnection services offered by the many ISPs and backbone providers co-located in the same facilities. Having access to premium peering and global interconnection services is critical for getting control over ping.
The internet and IP routing deliver traffic on a best effort basis, which means that by default they do not guarantee to deliver data or provide any quality of service. They also have several limitations that make consistently achieving low ping a challenge. There are multiple ways in which lag is introduced including buffering due to congestion, and buffers dropping packets when congestion persists. The small size of gaming packets — typically 55-bytes vs. 1,500 for the standard internet packet — makes for as much as 27 times the packet processing overhead for IP routers. The smaller packet size also results in more dropped packets because buffer limits are typically set using numbers of packets, not packet size.
Other lag or latency issues arise because of how IP networks calculate the routing of packets. BGP, which is the main routing protocol used in the internet, can create circuitous paths across the network, with more hops than necessary, and can even create different paths for egress and ingress traffic. As well, when it comes to peering or passing traffic to other networks, BGP does not consider the receiving network’s link capacity, real-time utilization, or network performance and may not select the same peering points for ingress and egress traffic. This can lead to very different performance results and high ping rates, particularly for gamers connected to different service provider networks.
There are several ways to overcome the deficiencies in these best-effort networks; most of them involve working around the deficiencies of IP and BGP. For instance, there are ways to engineer peering to ensure end-to-end gaming performance and enhance the gaming experience by moving traffic from congested peering links to lightly loaded links — or by selecting end-to-end paths with the required performance.
Peering engineering requires having an end-to-end view of the paths from the host server to the gamers that are using it. Thus, real-time network analytics are required to help ensure the best gaming experiences. By understanding gaming traffic patterns and how traffic traverses the internet and their networks, gaming companies can optimize end-to-end traffic flows between their gaming servers, content distribution networks (CDNs), the internet and gaming subscribers.
Sometimes, however, you need to build your own network and go around the internet, much as CDNs did a generation ago for video. Cloud gaming companies are increasingly implementing their own backbone networks to connect the edge and core cloud data centers that host their gaming servers. This enables much more granular end-to-end control, delivering the best gaming application performance. The internet in this case only acts as last-mile access for gamers.
Above: How gaming companies are building their own backbone networks to connect edge and core cloud data centers
Image Credit: Nokia
This new private interconnection fabric between data centers is usually composed of both routers and optical fiber trunks. To ensure that these links are deterministic — that is, they perform exactly as specified — both IP and optical layers of the network must be coordinated. In modern software-defined networking (SDN) that is the role of the central software controller. Analytics are built into the system so that, unlike with BGP, the performance of the end-to-end path, including peering points, is known and routers and optical links are set up by the controller according to specific performance policies and SLAs.
Beware of DDoS attacks
There is another dimension to the problem of network performance and gaming that relates to security. Just as online gamers are aware of ping, they are also very aware of denial-of-service (DoS) attacks. Hackers are constantly looking to launch cyber-attacks on popular gaming titles to disrupt games and target gamer accounts with fraudulent in-game purchases and microtransactions — or, sometimes, simply to slow a targeted gaming company down.
Distributed DoS (DDoS) is even more pernicious because hackers use botnets to attack from multiple hosts simultaneously. This makes it difficult for gaming companies to simply filter out a single attack point. Fortunately, end-to-end network analytics solutions are now capable of monitoring these multidimensional attacks, and they can direct specific edge routers to simply filter out attack-related packets from any point inside or outside the network, in real time.
Multidimensional network analytics and insights, peering engineering, and 360-degree DDoS protection can help radically improve the experience for online and cloud gamers. And with a private multilayer backbone network to connect edge and core data centers seamlessly with scalable, reliable, secure bandwidth and ultra-low latency, gaming companies can support massively scalable gaming services, meet the performance and latency needs of real-time games and deliver exceptional experiences for millions of gamers.
Jeff Sugimoto leads the global strategy and the technical sales teams for Nokia’s webscale business unit, and enjoys blending his 20-plus years of networking experience with a deep knowledge of cutting-edge technology to help customers build the next wave of internet services.
You can’t solo security COVID-19 game security report: Learn the latest attack trends in gaming. Access here
