Is WebRTC a Failure? - Kris Hopkins of CafeX

from Chris Hopkins Christa vegan come up so Chris has been working with Webber see technology like myself for a really long time he runs and leads product management strategy at one of the first commercial Weber see companies Cafe X I asked him to talk a little bit about to be honest with everyone here about whoever to see is it really that cool I mean I built the business promoting it but maybe I shouldn't be so Chris is gonna give us an honest opinion and talk about some of his experiences so please welcome Chris hoppin thanks everybody nice showing tonight as always thanks for hanging in there so honest feedback job all right so Dave Michaels this past week just said that WebRTC is a failure you guys can go home so the thing is that it's not a failure but it's definitely not perfect and we have a lot of experience at Cafe X deploying WebRTC in the enterprise and I think what what Dave Michaels is talking about the rough edges really the three main things that we encounter a lot with WebRTC when we go to an enterprise so Dave's an interesting guy but you know I don't think his expectation is that he thought WebRTC would remove all the phones on everybody's desk but he did think it would pervade into the enterprise a lot more and what we see is again the browsers that were forgotten we see some uptick you know here at Apple and I think that those are important notes to make but ie 11 even 10 Chris was talking about Comcast contact centers having ie6 still anyway we have and then when you start to go into multi-party conferencing so again if you don't want to use an MCU or a selective forwarding unit and you want to create a mesh conference you really peak out it around for concurrent parties in a conference right has anybody tried that anyway we'll get we'll talk more about that in a bit and then the architectures can get really complex so that browser situation where I II and Safari are being utilized is absolutely exacerbated inside the enterprise look at what internet explorer does so 75% 70% of the enterprises that we're in this particular survey showed that they use IE predominantly now we are seeing a move from the older versions of ie to ie 11 but that still doesn't help us with WebRTC for Cafe X we've had to go and create plugins to adjust to that and we also have plug-ins for Safari but as you know that's not an ideal user experience and of the top objections affiliate with WebRTC in terms of what's growing those same browsers ie and Safari are one of the major reasons why enterprises won't adopt it so they're waiting for something to happen now a lot of a lot of people will drag their feet just because they're saying you know look at the top the standards are incomplete but meanwhile we have how many applications are being used and utilizing WebRTC WebRTC despite the standards but I do think that the IE ins fari really our dragon feature and we encounter it every day so to wash this out and to see it through you kind of got to keep holding our breath right so there's a lot going on at Google Google is really pushing the limits Microsoft is starting to introduce ortc but simply an edge and then Apple we're seeing some we're seeing some uptick and anything Apple would do in terms of giving us access to the camera let's see in terms of get use user media would be you know welcomed we can just get started in a mesh conference really try it so when you try to go and have multiple parties talking to each other without a selective forwarding unit or without an MCU involved how you'll see what happens so I can get to about 6 on my laptop anybody else get higher have tried that no yes with YouTube broadcasting but real-time with everybody talking at the same time okay yeah and we see we see about four to five on a typical Haswell if you have a Core 2 Duo it's getting even lower Intel has made some claims that about 30% of the systems that are out there and in people's hands actually have processors that are three years or older so that means 30% of the population will peak out at about 2 to 3 parties when you're setting up again a mesh and you'll listen to it you can hear your fan will start going on you'll watch your battery drain when we first started doing WebRTC and we're at some of the first conferences I remember someone talking about an Android tablet they start doing multi-party on an Android and while the Android tablet was plugged in the battery would drain so so what happens is again we start to focus and this is where it holds true for cafe X there's a reason why we focused on contact center implementations with the WebRTC and that's because those tend to be a one on one engagement right you tend not to have multiple people talking to a customer service agent you definitely want multiple agents on the phone with a customer that's highly inefficient so that use case holds really true and it really helps us refine so again is the use case convenient to the technology or is the can technology convenient to the use case the other things that the architecture gets really complex quickly so if you take basic WebRTC you'll see the WebRTC triangle this is not the triangle of despair that Chris talked about this is the UM this is the triangle of hope but when Pierce can't connect inside the enterprise right because a firewall is variety of different things all of a sudden we start introducing five different server types and they all scale independently so imagine being me walking to a customer meeting and I'm talking to some poor procurement person an IT and I'm talking to the architect who's been there and he's had three vendor meetings prior to me and I'm starting to talk about a signaling server stun server a turn server media server all all when peers can't connect to each other and I'm pretty sure he turns to his buddy and he goes and says I'm I've been in three millions today and he just talked about beer not peers it's it's a lot to swallow for an enterprise again where the turn server resides that's often in the cloud they have they have reservations about using an open-source product they want something that's supported even though that that is a wonderful piece of software that's running it is a highly efficient they just don't quite trust it yet and then the architecture really gets this complicated so we went from the triangle of hope to the to the many beautiful pieces of software that we need to deploy inside the enterprise and again look look at this architecture so you have your web servers you have your stun and turn servers often those two are separated in some some architectures you need a media gateway right to do transcoding in other cases you need an actual media server to handle translating to stitch video together to uni a single for selective forwarding unit your signalling engine and they all scale independently so your signaling you know that can handle up to a thousand concurrent sessions in a single server but if you're doing transcoding you can get about two concurrent sessions per core on on that side these are all pretty standard industry metrics again some vendors are much better others are others are a little worse than that but it's hard and so anyway that's what we did so what I wanted to reassure you and again I'm keeping this quick because it is late and is keep after it it's nice to see so many of you dedicated to WebRTC we're all trying our best refine the use cases find those specific news cases that actually make a difference and you know we'll work keep the architectures less complex drive toward what you can and just keep after it we're all making this work and hopefully someday we'll prove Dave Michaels that he's not quite on the pulse of what's going on so anyway with that I appreciate your time and an adduct chat thank you for your brevity

Channel: WebRTC Boston

Views: 25,004

Keywords: webrtc, cafex, webrtc boston

Id: r-qIkkdGVLM

Length: 9min 21sec (561 seconds)

Published: Sat Mar 05 2016