Performance and Reliability: One Year Later

Growth from July 2012 to Present

Remember last Fall when we experienced an insane amount of growth during the back-to-school rush and experienced a few outages along the way? We do, and we’ve been working non-stop over the past year to make sure that we don’t run into the same issues.

We expect at least 2x the amount of traffic over last August/September, so we planned to 4x our capacity, but ended up engineering Poll Everywhere to handle 16x of anticipated capacity.

Not only did we scale our systems up by a factor of 16, but we also run a simulation of this traffic on our servers every night. That means while you’re sleeping, we’re busy hammering on Poll Everywhere to make sure its just as fast for you as the day before.

What does all of this mean?

It means you can expect fast, reliable service from Poll Everywhere this Fall and well into the future. Not only did we meet our scaling challenges, but we have introduced several new processes in our development and operations processes to make sure we always catch problems before they become a major issue.

But we didn’t just stop there…

We give you an unprecedented view into the status of our system once every 5 seconds

The live graphs at status.polleverywhere.com give you unprecedented transparency into our app to make sure everything is working if you suspect any issues.

Its pretty common for web applications these days to indicate the status of their system’s with a traffic light: green for “everything is OK”, yellow for “we’re experiencing some minor issues”, all the way to the dreaded red, “nothing is working!” That just wasn’t good enough for us, so we took transparency to the next level by providing these live graphs to our customers so they can see how long its taking for our system to process responses.

We run at least two of everything for increased reliability

We invested a lot of time and money to make sure that we run at least two of everything in our infrastructure to make our application more reliable. If a database server catastrophically fails, you probably wouldn’t even notice as it switches over to a backup system.

If that wasn’t enough, our customer support is legendary

Did you know that when you call our customer service number a human picks up that knows the product like the back of their hand? When you send us an email, its unusual if we don’t get back to you within an hour. When things go wrong or you’re having a tough time understanding something, we put the best support at your disposal. You can call, email, or hit us up on Twitter and we’re always there to help.

What other clicker or web response company do you know that commits to that level of transparency?