Participoll Goes Serverless
We’ve been working hard behind the scenes to boost the capacity and scalability of our interactive polling system for PowerPoint. Online polls create their own unique challenge for tech infrastructure. Normal websites have a steady flow of traffic with not much variation. Polls however are a bit more “feast and famine” with the potential to have large number of voters in a very small space of time with little or no activity in-between.
Up until recently we have been using server based technologies to handle our voting traffic with load-balancing and auto-scaling handling variations in traffic. However, auto-scaling is just not quick enough to react to really large spikes in demand so we have had to “bank” on maximum short term capacity. To that end we tested our previous set up to 3000 concurrent voters. But what about the challenge of really big polls? The solution is “server-less””
We have teamed up with AWS (Amazon Web Services) and their serverless technologies (Lambda, S3 and DynamoDB) to build an entirely demand based solution. We’ve done a ton of load tests, crunched our numbers and stopped counting at around 60,000 concurrent voters with still no drop in performance. All we need now is some hero presenters to challenge our new setup with some huge audiences.
Don’t forget that our standard license allows an unlimited number of voters, so whether you are presenting to 5 voters or 50,000 the cost is the same.