AWS Case Study: Envoy Media Group

Michael Taggart, Chief Technical Officer claims, “I will never build another web app on a traditional datacenter setup again. The fact that I never have to spend another night on a freezing datacenter floor fixing hardware has been a huge advantage. AWS has completely removed hardware support from the equation while still delivering on performance and reliability. It seemed like every trip to the datacenter would be at least a 12 hour ordeal. Many times I would head to the datacenter at 3PM only to arrive home at 4AM. Needless to say my wife is very happy with our move to AWS even though she doesn’t really know what AWS is.”

Envoy Media Group is an integrated direct marketing firm that offers solutions and highly-targeted media campaigns across multiple channels, including TV, radio, internet, direct video, and more. They currently host their entire web presence on AWS, including clusters of machines specifically dedicated to crunching algorithms related to multivariate testing, PPC bid optimization, and other compute intensive processes. The team also utilizes Amazon S3 and CloudFront. “We put CloudFront up against files served directly from the webserver and saw a noticeable increase in conversion when the CloudFront elements were being shown to visitors. The speed of CloudFront allows us to convert better and provide an improved visitor experience. This translates directly into dollars for us and we will be using it much more in the future.”

One of the key reasons Envoy moved to AWS was the ability to “pay as you go”. Taggart explains, “This pricing structure allows us to quickly and painlessly add servers where they are needed without large investments in hardware. The savings in time and man hours alone made AWS very compelling to us. In addition, we noticed a cost savings as we no longer have to maintain our own farm of hardware and can focus more on our apps and business model.”

“In a traditional datacenter setup, you tend to not think about server costs per hour so it can be difficult to compare the costs.” To explain the cost saving from moving from a co-location facility to AWS, Taggart says, “Our costs have been reduced by approximately 20%. We no longer have to pay for locker rental and power, our man hour costs have dramatically decreased, and our bandwidth costs are lower. For bandwidth, we had a contract for a 100 Mbps feed that was burstable to 1Gbps. The truth is our applications are not bandwidth intensive and we were only using about 1.2 Mbps of sustained bandwidth. With AWS, no minimums or contracts saves us immensely.”

Taggart continues, “AWS has enabled us to gain so much control over our server infrastructure. Before, we would make application design decisions around what hardware we had. Now, we are free to make our hardware work for us and whatever decision we decide would be best. It has also saved us numerous untold future costs. We are currently hiring at our company and can devote more resources to programmers and positions that have a direct impact on the bottom line instead of spending large amounts on a team of server admins. AWS really leverages the power of one admin, especially when used with services like Scalr, so that one admin can do the work of 10 or more in a traditional datacenter setup.”

Taggart offers additional advice to businesses thinking about using AWS, “First, I recommend anyone using AWS to use a service like Scalr.net or RightScale. It will save you so much time it is not even funny. We use Scalr.net and couldn’t be happier. The service only costs $100/month and Scalr will do things like automatically adding more servers when certain load thresholds have been reached, automatically replace servers that crash for whatever reason and automate pretty much anything you could possibly automate with AWS instances. In a word it is simply AWESOME. The second thing I would recommend is to test out AWS with your application before moving completely over to it. There are a few “gotchas” that can bite you in the butt if you are not ready for them. For instance, most developers assume that if they reboot a server, the files that were there will still be there when the server comes back up. With AWS this is not the case unless you use EBS volumes, backup to S3, or use brew your own networked file system. We actually use a combination of all 3 to make sure our data is persistent, reliable, and fast. We currently use GlusterFS as a networked file system for webservers and workhorse servers, EBS volumes for DB servers and Gluster storage nodes, and S3 for backups and other storage.”

“To summarize, AWS is the superhero of my tech team and AWS really stands for Amazing Web Service.”

To learn more about Envoy, visit http://www.envoymediagroup.com

(Published January 2010)