Where have all the CDN-disrupting technologies been this year, while traditional server-based networks have been resolutely handling the relentless demand for OTT video the world over?
Last year was a story of peer to peer (P2P) networks creeping out of the shadows – spearheaded by two market-altering deals with Liberty Global embracing Teltoo and CenturyLink acquiring Streamroot. We have even witnessed the emergence of blockchain-based overlay networks to further make the established CDN vendors sweat. This year was therefore tipped to be the year that decentralized P2P software finally entered the mainstream, but along came lockdowns, and suddenly the resilience of the classic CDN was solidified – as demonstrated during the emphatic Akamai Edge Live event earlier this year (coping with traffic spikes of as much as 85% in some early lockdown regions).
The P2P community remains optimistic, however, with German start-up StriveCast expecting demand to go through the roof next year when the budgets of target customers are taken off ice. After all, who wouldn’t stop and stare at a poster promising 75% CDN bandwidth savings?
“Stakes are insanely high right now – it’s the perfect time to position ourselves,” said StriveCast co-founder and CEO Alexander Schaefer, speaking to Faultline this week.
In fact, contrary to what the German P2P start-up’s website might say, Schaefer prefers to describe his company as an enterprise CDN (eCDN) provider, rather than a classic P2P peddler. Initially, we thought this change in tone was perhaps due to the historical piracy connotations associated with P2P technology, but rather the term eCDN is to distinguish from P2P networks used alongside direct to consumer CDNs, with Schaefer citing Microsoft Teams as probably StriveCast’s biggest customer. It also has an operator scalp under its belt with Swisscom, while Kaltura is a familiar name in the Faultline ecosystem making the most of StriveCast’s eCDN.
So, in that sense, StriveCast is helping offload the freakish demand for services like video conferencing – making massive CDN bandwidth savings on symmetrical video delivery and freeing up bandwidth for other applications like OTT video in the process. The Robin Hood of 2020’s bandwidth-crunch is probably a stretch, but you get the picture.
“We are seeing six-figure requests for eCDNs. Covid-19 has highlighted the need for symmetrical video,” he added.
“OTT broadcast is not so simple. We are mostly focused on enterprise, but there is still some focus on the OTT broadcast market. We have not given up on broadcast OTT, but with live events cancelled, it has allowed us to really take a step back,” said Schaefer, responding to our comments on the notoriety achieved by Teltoo and Streamroot.
As for what makes StriveCast unique technologically, that is difficult to comprehend without a wall of computer science degrees, because servers within P2P networks are fundamentally the same, made up of client devices that form a software-defined overlay based on the open source WebRTC framework. “We are not trying to reinvent the wheel here, but the details really do matter. The difference to a company like Teltoo is that we can mathematically prove how we can be better, how we can endure it and how you can prove it works with no buffering. StriveCast’s fundamental difference is how the software is managed, and you always need a bit of latency,” admitted Schaefer, not giving too much away of his company’s secret sauce.
In other words, StriveCast is a server-side managed P2P network – creating a dynamically growing, distributed delivery network between all the connected devices streaming your video content.
A common drawback of P2P networks is having an adverse impact on ISPs, shifting burden of capacity provision from the content providers and CDNs to the ISP network, which was largely the case with BitTorrent. For this, StriveCast’s OTT product includes a fallback feature that automatically detects if devices are struggling to send or receive data, and instantly switches from P2P to the underlying CDN.
This caused us to think back to a conversation with Limelight Networks earlier this year, when we were surprised to hear the US CDN provider speak openly about embracing P2P architectures, describing the software as complementary to its own web of servers. Schaefer was equally surprised by Limelight’s P2P nod, saying this is not something he is used to hearing from the CDN community.
“With a P2P network, you drastically reduce traffic with fewer requests. Akamai engineers have a heart attack when they see this kind of thing, because a P2P network cannot interfere with the CDN,” he added.
That said, hybrid CDNs (sometimes called peer-assisted CDNs) have evolved, which we anticipate becoming a major direction of travel not just for start-ups but also mainstream CDN providers for video delivery. That is because the bandwidth savings enabled by such peer assistance in the face of relentless growth in the volume of video traffic, especially over mobile networks and in developing markets, will be crucial. This will be driven by start-ups that have recognized the need to add the QoS mechanisms required for video streaming that have been lacking in taboo P2P networks like BitTorrent, typically using the WebRTC protocol as that is inherently P2P in its operation and is also optimized for ultra-low latency.
Founded in 2017 by computer science graduates who spent five years researching the subject, StriveCast’s first PoC demonstrated 200 viewers on Twitch – proving an impressive 75% bandwidth savings, slightly less than the 85% savings promised by Teltoo, which is what swayed Liberty Global.
Our skepticism of these high savings figures to be achieved consistently on a large scale in major deployments was met with an air of familiarity by Schaefer. “75% bandwidth savings is something we see commonly,” he assured us. However, just as important as the savings rate KPI is the utilization rate KPI, which pertains to the percentage of the audience that can actually partake in a P2P network. For example, you must have UDP enabled and be able to stream WebRTC, so it can be the case that only 5% or 10% of peers in a network are available to participate.
Scahefer pointed to Hive, Collective and Ramp as companies using P2P technology for enterprise customers, specifically for internal delivery, say a CEO wants to make an announcement to thousands of employees within the corporate network by peering each device. So, sort of like a P2P intranet.
Ending on a dubious note, another criticism we have heard from the traditional client/server architecture using a one-to-one unicast basis, is that P2P networks typically do not dovetail well with the CDN orchestration layer, which was not nearly specific enough for Schaefer’s liking. “When you manage thousands of PoPs and caches, of course lots of orchestration comes with you,” was his frank answer to that conundrum.
While we are hesitant to coin 2021 the year that P2P networks become the “new normal” – to take a leaf from the pandemic book of buzzwords – certainly a wave of new start-ups like StriveCast will be instrumental in driving decentralized software deeper into the architectures of major video streaming platforms that have almost become a commodity, just as the traditional CDN has.