There were two disappointments for me from last week's WebRTC conference in Paris & recent ongoing news:
a) Still no major public WebRTC involvement from other big Web players beyond Google (unless you count Amazon Mayday which is still unconfirmed as WebRTC or not)
b) No really exciting new things happening in WebRTC DataChannel for the past 6 months
Both changed yesterday, with Yahoo! buying WebRTC startup PeerCDN.
In my presentation at the Paris conference, I predicted that 2014 (see Slide 22 here) would see "major Internet players using WebRTC" and "Surprises, especially around data use of WebRTC". Looks like even I've been caught out by the speed of WebRTC evolution - they've both happened with 2 weeks of 2013 still left.
Silicon Valley jokes about Yahoo! acquisitions aside, this is a really interesting move. PeerCDN is one of a handful of startups trying to use WebRTC's browser-to-browser data communications ability, to reduce bandwidth/CDN costs for content companies - and ideally to reduce latency as well. Peer5 and StreamRoot (which was in Paris last week) are also in this space.
The idea is simple - rather than having all users download content from the company's own servers, or that of a commercial CDN like Akamai or Limelight, have some of it downloaded from each other. If two people are on the same site at the same time (or perhaps with cached content in their browsers), connect the two browsers together using WebRTC's less well-known DataChannel API, and deliver certain chunks of content like that instead.
There are various implications:
- Reduce the outbound bandwidth costs (or CDN delivery costs) of the content company
- Improves website response times, especially if the users concerned are geographically close to each other
- Indirectly increases uplink local-access bandwidth consumption, as the users transmit some of their data back "upstream" to the other peers
- Represents a legal use-case for P2P data, as the content company is in control, and indeed in approval.
- Impacts the potential revenues and margins for traditional CDN players, and massively reduces entry barriers to that market
- Makes it harder for telcos to monetise on-net CDNs - although perhaps easier to launch them, if they white-label services such as StreamRoot's.
- It doesn't matter if only a % of browsers are WebRTC-capable. Where it's not supported by a given user's device, the content delivery reverts to Plan A - some reduction in bandwidth costs is definitely better than nothing.
Now, it's one thing doing all this on a small scale and demonstrating it at WebRTC conferences. It's quite another building it into a service that competes meaningfully with the likes of Akamai, which have global footprints, established salesforces and customer relationships, and many other value-added services - and also a fearsome army of IPR lawyers and an arsenal of patents.
Yahoo!, for all its problems, is at least in the category of companies that can play those games - or else perhaps it may prefer to use the technology internally, rather than exposing it as a 3rd party service. Various commentators have suggested it might speed up its notoriously laggy webmail, but I don't think that's the key application here, as little content is shared between users. It's also possible that the CDN project will get lost in the background, with Yahoo! actually just buying the three engineering innovators behind the work, for more general development on WebRTC.
As for DataChannel, this is one more signal that it's perhaps even more important and disruptive than the voice and video bits of WebRTC. We've already seen it used for various types of screen-sharing, Google's Chromecast dongle and a few other early attempts. If the CDN use-case proves properly commercialisable by Yahoo/PeerCDN or others, then it's another landmark.
Other use-cases for DataChannel are also being widely talked about - certainly app/screen-sharing will be a major (if unspectacular) one, most likely bolted onto conferencing and UC solutions. But the real "fun stuff" will occur when it enables realtime data streaming for things like sensors, Internet-of-Things objects, health monitoring and so forth. Let's see what happens - but I have a feeling that we'll get a lot more than just yet another "Skype in the Browser" clone from DataChannel.
This is also an area that telcos should pay attention to - well away from the ongoing IMS/WebRTC integration for VoLTE or anything else. I've said for a while that WebRTC needs to be dispersed across multiple business units in operators, and not just left to the voice/core-networks group. Opportunities for exploiting WebRTC DataChannel should be high on the priority lists of the CDN group, content/video optimisation, M2M, healthcare & other verticals, devices and numerous other telco business units. If operators don't get going on this immediately - at least in the labs or other skunkworks - then who knows what other 3-person startups will create independently.
Overall, I think this acquisition is a very interesting move. It marks the first overt buyout of a WebRTC player by a Silicon Valley behemoth (Telefonica/Tokbox was something else). That could trigger more VC attention, as well as focus attention on whether Google or Apple or Facebook or LinkedIn also start spending money on the next stage of the web. It's also a wakeup call for all those in the WebRTC community who forget that data is often at least as important as voice or video.
Interested in reading more analysis & comment about WebRTC? I am the only analyst who has done a full, cross-sector research report, covering all major use-cases, with conclusions and forecasts updated regularly. Click here for details of the Disruptive Analysis WebRTC report & update subscription.
a) Still no major public WebRTC involvement from other big Web players beyond Google (unless you count Amazon Mayday which is still unconfirmed as WebRTC or not)
b) No really exciting new things happening in WebRTC DataChannel for the past 6 months
Both changed yesterday, with Yahoo! buying WebRTC startup PeerCDN.
In my presentation at the Paris conference, I predicted that 2014 (see Slide 22 here) would see "major Internet players using WebRTC" and "Surprises, especially around data use of WebRTC". Looks like even I've been caught out by the speed of WebRTC evolution - they've both happened with 2 weeks of 2013 still left.
Silicon Valley jokes about Yahoo! acquisitions aside, this is a really interesting move. PeerCDN is one of a handful of startups trying to use WebRTC's browser-to-browser data communications ability, to reduce bandwidth/CDN costs for content companies - and ideally to reduce latency as well. Peer5 and StreamRoot (which was in Paris last week) are also in this space.
The idea is simple - rather than having all users download content from the company's own servers, or that of a commercial CDN like Akamai or Limelight, have some of it downloaded from each other. If two people are on the same site at the same time (or perhaps with cached content in their browsers), connect the two browsers together using WebRTC's less well-known DataChannel API, and deliver certain chunks of content like that instead.
There are various implications:
- Reduce the outbound bandwidth costs (or CDN delivery costs) of the content company
- Improves website response times, especially if the users concerned are geographically close to each other
- Indirectly increases uplink local-access bandwidth consumption, as the users transmit some of their data back "upstream" to the other peers
- Represents a legal use-case for P2P data, as the content company is in control, and indeed in approval.
- Impacts the potential revenues and margins for traditional CDN players, and massively reduces entry barriers to that market
- Makes it harder for telcos to monetise on-net CDNs - although perhaps easier to launch them, if they white-label services such as StreamRoot's.
- It doesn't matter if only a % of browsers are WebRTC-capable. Where it's not supported by a given user's device, the content delivery reverts to Plan A - some reduction in bandwidth costs is definitely better than nothing.
Now, it's one thing doing all this on a small scale and demonstrating it at WebRTC conferences. It's quite another building it into a service that competes meaningfully with the likes of Akamai, which have global footprints, established salesforces and customer relationships, and many other value-added services - and also a fearsome army of IPR lawyers and an arsenal of patents.
Yahoo!, for all its problems, is at least in the category of companies that can play those games - or else perhaps it may prefer to use the technology internally, rather than exposing it as a 3rd party service. Various commentators have suggested it might speed up its notoriously laggy webmail, but I don't think that's the key application here, as little content is shared between users. It's also possible that the CDN project will get lost in the background, with Yahoo! actually just buying the three engineering innovators behind the work, for more general development on WebRTC.
As for DataChannel, this is one more signal that it's perhaps even more important and disruptive than the voice and video bits of WebRTC. We've already seen it used for various types of screen-sharing, Google's Chromecast dongle and a few other early attempts. If the CDN use-case proves properly commercialisable by Yahoo/PeerCDN or others, then it's another landmark.
Other use-cases for DataChannel are also being widely talked about - certainly app/screen-sharing will be a major (if unspectacular) one, most likely bolted onto conferencing and UC solutions. But the real "fun stuff" will occur when it enables realtime data streaming for things like sensors, Internet-of-Things objects, health monitoring and so forth. Let's see what happens - but I have a feeling that we'll get a lot more than just yet another "Skype in the Browser" clone from DataChannel.
This is also an area that telcos should pay attention to - well away from the ongoing IMS/WebRTC integration for VoLTE or anything else. I've said for a while that WebRTC needs to be dispersed across multiple business units in operators, and not just left to the voice/core-networks group. Opportunities for exploiting WebRTC DataChannel should be high on the priority lists of the CDN group, content/video optimisation, M2M, healthcare & other verticals, devices and numerous other telco business units. If operators don't get going on this immediately - at least in the labs or other skunkworks - then who knows what other 3-person startups will create independently.
Overall, I think this acquisition is a very interesting move. It marks the first overt buyout of a WebRTC player by a Silicon Valley behemoth (Telefonica/Tokbox was something else). That could trigger more VC attention, as well as focus attention on whether Google or Apple or Facebook or LinkedIn also start spending money on the next stage of the web. It's also a wakeup call for all those in the WebRTC community who forget that data is often at least as important as voice or video.
Interested in reading more analysis & comment about WebRTC? I am the only analyst who has done a full, cross-sector research report, covering all major use-cases, with conclusions and forecasts updated regularly. Click here for details of the Disruptive Analysis WebRTC report & update subscription.
No comments:
Post a Comment