The point behind Ernie is that we need a framework to create server based, presence and VoIP enabled applications for next generation service providers and enterprises. Our first applications should be able to highlight these uses. After some internal debates, here's a list of Ernie's first applications, given in order :
1) An automated testing application for service providers. Ernie servers will sit on the edges and the boundaries of service provider networks, and will provide an active testing platform for them. Probably forward testing information to a Nagios system for display, monitoring, etc. We'll make it work with analog phone adapters as well, so you can test complete IP to PSTN connectivity. Ernie will test completion rates, voice quality, etc.
2) A basic contact center application that will handle inbound calls, agent interfaces, queing, etc in a SIP based network.
3) A find-me, follow-me service that can integrated into web pages. Presence and VoIP enabled.
Wednesday, April 12, 2006
Hi all -
So, now that I'm all married and stuff, let's get back to the grindstone.
Here's some prinicples of Ernie's architecture and design:
1) It should be standards based, as much as it can be.
2) The components should be naturally distributed and scalable. For instance, we should be able to add in a Convedia or Snowshore media server to accelerate and enhance media processing.
3) As much as possible, it should be fully redundant and fault tolerant.
4) It should be multi-application and multi-user friendly. Someone should be able to login to an Ernie server and manage (start, restart, stop, check, pause...) the applications that are running there.
5) People may want to make money with the Ernie applications they have written. Billing interfaces should be built into the platform.
6) The highest interface should be as simple as possible. Simple things should be simple to do. The ultimate customers are the millions of web programmers.