Cinchcast uses the cloud to scale up its self-serve conference calls

Cinchcast lets you create your own massive webcasts and conference calls for events such as quarterly conference calls. The number of participants on the calls can scale from a handful to millions of people. Without the cloud, this wouldn’t be possible.

[aditude-amp id="flyingcarpet" targeting='{"env":"staging","page_type":"article","post_id":581470,"post_type":"story","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"big-data,business,cloud,","session":"C"}']

Aleksandr Yampolskiy (pictured right), chief technology officer at New York-based Cinchcast described the key lessons from scaling up Cinchcast using cloud services at VentureBeat’s CloudBeat 2012 conference in a fireside chat with Vanessa Alvarez (pictured left), director of product marketing at Gridstore. Cinchcast disrupted the conference call business by creating self-serve, reservation-less web-based conferences.

“We scale from hundreds to tens of millions,” Yampolskiy said. “With the influx of traffic, you can easily provision extra instances” of server hardware.

Yampolskiy said he relies on three rules to help scale up the company’s business. The first rule is to automate as much as possible. When a process requires an engineer to work until 2 am in the middle of the night, it clearly isn’t scalable or automated. Rule No. 2 is simplicity. He says you have to make complicated things simpler in order to scale a service to the millions. Rule No. 3 is to use metrics to measure every aspect of performance, security, uptime, and latency.

Those rules were put to the test on June 29, when usage of the site began to plummet. Over the course of a few weeks, traffic fell from 4.8 million page views a day to 4.2 million. Executives were camped outside Yampolskiy’s door seeking an explanation. His company used New Relic‘s performance management app as an X-ray for the business. The team used it to identify a problem known as “donut caching” that wasn’t working properly, resulting in a slowdown in performance for users. The company attacked the problem and improved the service.

Since Yampolskiy joined 15 months ago, the load time for the site has gone from 14 seconds to 4.9 seconds.

“We saw a huge corresponding increase in revenues,” he said.

As for a public or private cloud, Yampolskiy said, “I like to be in control of my destiny. If it goes down, it is my responsibility. But I have a mix of public and private clouds.”

Photo credit: Michael O’Donnell

[aditude-amp id="medium1" targeting='{"env":"staging","page_type":"article","post_id":581470,"post_type":"story","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"big-data,business,cloud,","session":"C"}']