- Update
2017-11-20
- Author
- Author's site
- version
0.5.0
- Session Date and Time
2017-11-20 21:35-21:45
- Session location
TECH PLAY SHIBUYA
- Presentation time
10 minutes
This document is the presenter's material of WebRTC Meetup Tokyo # 17 Special Edition conducted in 2017-11-20 . For questions on this material, please mention to @voluntas on Twitter. (I can receive DM from everyone.)
I’d like to talk about the possibility whether we can run a business successfully with WebRTC product.
Possible.
However, we are a small company with almost only engineers and we don’t have marketing or sales team. If you are interested, please read the articles below.
- Now we can cover the salary for our employees by the sales of our own product.
- The beggining of our own product
I am an engineer mainly using Erlang/OTP at a company called Shiguredo. I usually write code for our product or provide technical consulting for various enterprises such as game developing companies.
We decided to make a “Package Product”. We thought it would be difficult to make a profit as a WebRTC “Service provider”.
- Package product with closed source
- Make a profit from annual license fee
- Affordable fee even for small enterprises
We think WebRTC is great technology anyone can use easily, so we wanted to set our license fee to as low as we could, which allowed small enterprises possible to pay.
Basically WebRTC Service model has only two patterns.
- A model that provides a signaling server including a TURN server
- A model providing MCU or SFU
In the former model, there is no choice but to make a profit by maintaining and operating signaling server and TURN server. Those servers can now be relatively easy to set up even personally. In other words, it is quite a difficult model to make a profit for us. In addition, SkyWay is offering free of charge of TURN, up to a certain amount, it must be hard to compete with them.
Since the latter MCU/SFU offering model is necessarily via the server, problems of forwarding data traffic will occur. The more customers we have, the more traffic we use. If we charge our customers according to their traffic amount, the profit will be small as long as we ourselves also need to pay transfer fee.
Therefore, we decided not to be a Service Provider but to be a Package maker. A further talk will be off the topic, let me skip the details.
Actually, I am an OSS person, but we decided to make our own product with closed source. I know well from my own experience that many companies in Japan are reluctant to adopt Open Source. Most enterprises that understand the value of Open Source and pay its support fee are large companies, and they also buy products with Closed Source as well.
Only two things that we focus on
- Simple (minimal function)
- Tolerant
Our product can only “distribute” and “record”, and so far we’ve never received a report from our customers that the server has gone down.
WebRTC is a technology used for exchanging voice and video, that means it is very easy for end users (=our customers’ customers) to notice when server goes down . Therefore, it is crucial for us to provide our customers with a tolerant, and easy-to-use product.
In order to sell our package well, we think of SDK as important as our product itself.
We have three types of SDK for our product, JavaScript, iOS, and Android. All of these are provided as Open Source under the Apache License 2.0.
We think the quality of SDK is one of the important keys for our business. We are investing a lot of human resources relatively to make our SDKs better, and we are actively updating them.
Since we have been following the latest version of libwebrtc, we have accumulated its Know-how. Our CTO wrote the following article.
WebRTC Native iOS Library API Guide
Next year, we are planning to add functions to convert or synthesize video and audio with utilizing GPGPU. That is, “MCU” function. We think a product that combines MCU and SFU is necessary to compete advantageously in the future WebRTC market.
In addition, we are considering maintaining WebRTC library for React-Native by ourselves.
In order to run a business with WebRTC product successfully, it is essential for us that our customers make their service successful with our product.
Therefore, the most important thing is we continue to provide a stable product for them to make their service more popular.
It is a commercial WebRTC SFU. The annual license fee is 600,000 yen per 100 concurrent connections.The license fee includes support fee. It can be used for various purposes such as conference with multiple people, distribution to hundreds of people, one-on-one interview. Since it is offered as a package, it can be operated in-house. Whether AWS or GCP, on-premises or anything else. If only you have a server, it will take less than 10 minutes to install and our product. Furthermore, since our product has built-in demo function, 15 minutes are enough to experience our live communication.