Developing a WebRTC service? Amazon Web Services is the best place for it.
When I’ve written the piece last week about AWS being a bad hosting choice for WebRTC, I assumed it would bring in some comments, and it did. I also indicated the fact that I am not sure myself what the answer should be. So here’s the other side of the same coin
You looking to place your new WebRTC service somewhere? My suggestion? Skip GoDaddy. And while at it, don’t build your own data center or go to a specialized provider. Use the best that the cloud has to offer – go for AWS.
Why? Here are a couple of reasons:
- You need media and signaling. Signaling will be the bulk of the logic in your application and media will be the bulk of the bandwidth and CPU usage. This being the case, you will either need to split them between providers or just host them under the same roof. AWS is a great roof for such variety
- Scalability, reliability and global footprint. AWS has it all. Today. In many locations around the globe. With proven vendors running there already. And an API that enables you to automate everything about how you use it
- Dedicated servers. Amazon has dedicated servers. We can now argue if these are really dedicated or just a trick in virtualization, but best would be to just check and see for yourself. They might be just the thing you need for predictability in your media processing
- Amazon Elastic Transcoding. There is such a service, you know? It can take WebM (=VP8) or H.264 and transcode them. Which works nicely for archiving and broadcasting on a streamed channel
- Richness. Come to think of it, AWS isn’t just a compute capability. It offers higher abstraction levels of storage, database, NoSQL, messaging, email, etc. It makes it so much easier to develop services on top of it. Check out Beanstalk…
- Price. Did you know that Amazon just slashed their prices? Again? They do that all the time. You won’t be stuck with high costs on old resources
If you go with AWS, it will probably be easier to find developers and DevOps guys who are knowledgeable with it. it will get you faster to a demo and the minimum viable product. If you won’t like the performance, you can invest the effort to switch once you raise more capital – this isn’t the catholic church.
What’s there not to love about such a solution?