Takeoff is a very cool web app that helps people collaborate on video. They recently published a great blog post outlining their experience with video encoding. It’s worth a read if you’ve ever worked with a video encoding system, or ever will.
A few interesting bits:
They started by rolling their own transcoder on AWS, but quickly realized that this wasn’t the best approach: it was slow (20 minutes for a 1 minute video!), and wasn’t set to scale beyond a single server.
Over the next year Takeoff grew. The response from our users was overwhelmingly positive. They loved the app. But they weren’t too fond of our encoding. Complaints rolled in about long wait times, a video never encoding, etc. We needed a change, so we took a fresh look at cloud encoding platforms.
They decided to try Zencoder, and were able to make the switch within just a few hours.
So we decided to make the switch. I started coding the integration at 8:30 am the next day. By 2pm I had the whole system running off their platform. We tested until about 8pm, throwing everything we could at Zencoder. Not a single issue, so we deployed the update and went to bed with huge smiles.
Finally, we’re happy to say that we were able to speed up their video encoding by 10x.
Two minutes later I opened up the helpful Zencoder dashboard to see the job status. It said it was done. Whaaaat? That same video took 20 minutes on our homemade encoder. How could Zencoder do it that fast? There had to be a problem, but nope, no error. A 30mb mp4 was waiting for me. It was true – Zencoder was 10x faster than our system.
This is why we come to work in the morning. We don’t just want to write software; we want to write software that customers love, that makes their lives easier, and that makes their products better.
Building software is fun; building awesome software is fulfilling.