AEG, I am a solutions architect with focus on unified communications, web 2.0 and telecom equipments/applications. I suggest or use Plivo Open Source as much as I use other platforms like Adhearsion, Telegraph (sometimes Twilio, Tropo, Mayo, Aculabs, Callfire and many others) as needed. Or I simply do some development on FreeSWITCH/Asterisk with my team to handle custom needs and then testing, benchmarking, support...
Having said that, tell us exactly what you need to achieve using Plivo and I will try to answer (or we will seek an answer from more experienced group members Mike or Venky).
To answer your question in the 2nd paragraph of your email, try the following:
Suggestion 1: Use DNSSrv for setting up a failover 'availability zone' on AWS for highly available frontend. Setup failover acoss various AWS regions.
Use 2 OpenSIPS instances (Active - Passive) and setup HA configuration and load balancing for all FreeSWITCH instances you want to run. Any custom development can be done on OpenSIPS in this case.
Suggestion 2: In the same setup as as above use AWS Elastic Load Balancer with a custom script which launches a machine image with freeswitch and all pre-requisites as soon as the load on another freeswitch instance reaches specified threshold and modifies the openSIPS load-balancing module setting to include the newly launched FS instance in the LB cluster. In this way AWS will handle load balancing and you may run very large volume of calls with auto scaling.
Suggestion 3: Use FreeSWITCH as SBC and OpenSER (not sure about FreeSWITCH's LCR module hence suggesting OpenSER for high volumes and huge routing tables) as the routing engine. You can setup AWS ELB script to load balance many Plivo+FS instances or one Plivo with many FS instances (not tested).
You will need to test in your requirement scenario and may be able to use some of the traditional telephony equipments which can not be virtualized as well as some billing server, provisioning server, RADIUS etc virtualized using OpenStack.
Regards