Avoiding single point of failure in a multi-services architecture
The audience would learn how we at Plivo avoid single point of failures, in a complex architecture that has lots of moving parts and is on the cloud as well as dedicated hardware.
We at Plivo provide a platform for developers to integrate telephony into their applications by abstracting away the core plumbing and exposing it over an easy to use REST API.
Being in this space, high availability is of paramount importance. Our users build apps that end up being used by various people who may not be very technical. For them, the requirement is pretty simple, the phone line should never go down!
The backend of such a product is complex to say the least, with lots of moving parts in the architecture all having an independent function. These moving parts are then glued together to provide the service as our users see it. In this talk I aim to discuss how we ensure high availbility by making each component HA, which results in keeping the service highly available.
I’ve been working with open source telephony for over 5 years. Starting with pure telephony functions like hardware PBX systems to gradually moving towards web telephony that encompasses more than just SIP and RTP.