A sudden and dramatic surge in internet traffic, much of which is driven by video streaming and game and software download events, is testing network infrastructures with unprecedented volume. The 30% increase in global traffic Akamai delivered from the edge during the past month, compared to the ~3% growth during a typical month, illustrates the impact COVID-19 has had on the internet. As a result, regulators, ISPs, and many Akamai customers are taking action to reduce load during peak traffic times to avoid excessive congestion.
Akamai is collaborating with our customers, who are working to do their part to reduce bandwidth usage during peak times. These efforts are helping to ensure everyone is able to continue enjoying great online experiences despite historically high traffic levels.
A Surge in Video Traffic
It's hard to fathom another time when more collective eyeballs were in search of something to watch. During the coronavirus pandemic, streaming services have emerged as a leading source for entertainment and information. That's led to skyrocketing viewership along with a need to optimize bandwidth consumption, especially during extended peak time windows in regions that are seeing excessive growth in internet usage.
In countries or regions in which regulatory bodies are providing guidelines for bandwidth management, Akamai recommends that customers adhere to those. In many cases, reducing bitrates doesn't necessarily mean reducing video quality. Often, bitrates can be lowered without impacting perceptual quality. This can be measured by using perceptual quality assessment tools. To learn more about these tools and how to apply them, check out this research paper. Here are a few recommendations on how customers can go about optimizing their video bitrates for peak times.
Optimizing Bitrates During Preparation
Content providers can use context-aware encoding solutions during the preparation process to optimize the bitrate ladders for each title in a catalog. Establishing a ladder for each title sheds unnecessary bandwidth usage without affecting the perceptual quality of video content.
Employing high-efficiency codecs is another way to optimize for lower-bandwidth video delivery at the same or similar perceptual quality. For example, HEVC offers an approximately 50% improvement in compression efficiency compared to AVC. VP9, AV1, and a number of other codecs also warrant consideration. When choosing codecs, be sure to identify which ones require royalty fees, and select the combination that will support the necessary range of target viewing devices.
Optimizing Bitrates During Delivery
Another approach to reducing bandwidth during peak times is through manifest manipulation. This offers content providers the flexibility to keep their current video workflows intact, and edit video manifests with customized bitrate ladders at the edge, just prior to delivering to an end user's device. During peak times, manifest manipulation can remove the top bitrate renditions across the board or target particular geographies in which to limit bitrates.
For those interested in manifest manipulation, Akamai is offering 30 days free use of Adaptive Media Delivery to help customers deliver high-quality video at scale. The offer includes 30 free days of the Manifest Personalization option to personalize manifests and deliver lower bitrates from the edge. Manifest personalization is currently available for both HLS and DASH.
Optimizing Bitrates Using Player Intelligence
Content providers can also use the advanced logic built into a video player to augment bitrates. A wide range of players are proficient in selecting appropriate streams based on factors such as screen resolution, network condition, and device capabilities. Many player solutions include the ability to cap bitrates at a certain threshold and override the player logic.
Akamai's Adaptive Media Player (AMP) can be configured to limit bitrates to reduce bandwidth usage across all of its platforms. For AMP Web, see the configurations for delegated libraries for HLS here and DASH content here. For AMP iOS and Android SDK, consult the iOS docs for setMaxQualityLevel() and the Android docs for setMaxBitrate(). For those interested in testing this capability, Akamai is offering 30 days free use of Adaptive Media Player.
Surging Download Traffic
Planning Around Peak Times
Software downloads, -- whether for video games, patches, or system updates -- are among the biggest traffic drivers on the internet. Shifting high-traffic download events to off-peak times can help reduce spikes and throughput, and the subsequent congestion. This can help ensure that event-related peaks don't coincide with regular traffic peaks, helping networks to operate within their normal capacity thresholds. Further steps can be taken through staggering downloads in batches by geography and/or platform wherever possible. Software patches, game updates, and other schedulable events should be rolled out during off-peak times for their respective geographies -- generally 12 AM to 7 AM local time. In some instances, updates can be pre-positioned or pushed to users incrementally during off-peak hours, with the push spread over several days.
Game developers and console manufacturers can work together to coordinate large releases, limiting the overlap and minimizing traffic peaks created by concurrent releases. This type of collaboration can help serve the greater good of the industry and all of its end users.
Optimizing Throughput During Delivery
Those looking to deliver downloads during peak times but limit the speed of the download can do so by working with their CDN provider to set a threshold for throughput during periods of congestion. Akamai's Download Delivery includes a max pacing rate setting that allows users to control the rate at which the download occurs.
For those interested in working with their CDN partner to accomplish this, Akamai is offering 30 days free use of Download Delivery to help customers deliver high-performance downloads at scale, with the option to control download speeds during certain times.
Optimizing Throughput Using Client Intelligence
Software and game publishers who manage software on the device initiating the download can also utilize client intelligence to limit bandwidth when necessary. In fact, the simplest and most efficient way to achieve this is often at the client. Recently, this approach has gained traction as several publishers have released client updates that limit throughput for downloads or give end users the option to do so.
During his recent Edge Live presentation, Akamai CEO Tom Leighton said, "I've been looking at traffic graphs for over 20 years now, and I can't recall seeing anything like this."
These truly are unprecedented times. It's up to all of us, as good internet citizens, to do our part in helping to ensure these critical lines of communication and commerce remain open and unimpeded.