The Speed the Cloud winner from Interop

At Interop Las Vegas, this past Thursday afternoon at just after 2pm, we ran the Speed the Cloud contest demonstration.  What that means is that we performed the actual live file transfer in front of a very curious and interested audience, most of whom had entered the contest.

The demonstration consisted of two official file transfers.  We placed a 50MB PowerPoint file on a cloud server in Virginia, and we pulled that file to our booth in Las Vegas across the public Internet.  The link was a T1 (1.5mb/second), and the connection to Virginia had about 90 ms of latency.  To enter the contest, a user had to use Twitter, and:

1. Follow @riverbed on Twitter
2. Generate a single tweet containing guesses for the time it would take to complete both the unoptimized transfer and the optimized transfer with the hashtag #speedthecloud

The entry whose guesses were the closest to the actual times would win a $500 Apple Gift Card. 

First we transferred the file unoptimized.  The transfer took 4 minutes and 51 seconds, or 291 seconds.  50 MB in 291 seconds comes out to 171 kilobytes per second. 

Then we brought up the optimized connection and began that transfer.  After a few seconds of transfer, it was clear that the file was cold.  That means that the Steelhead appliances involved in the transfer had not seen the file before.  Our rules said that the transfer would be warm, where the Steelheads HAD seen the files before, so we used this transfer to warm the file, but it did not count in the final results.  The cold transfer took 2 minutes and 21 seconds, or 141 seconds.  That's 354 kb/second, or a little better than double the performance of the unoptimized transfer.  But again, just to be clear, that doesn't count toward the contest.

Then when that was done, we started the 3rd and final transfer, the one that would count toward the contest.  Since this was a warm transfer, we knew it would be pretty quick.  It completed in just 4.1 seconds, which is a remarkable 12.19 megabytes per second.  Not bad for a T1.

So here's a quick graphic comparison of the three transfers:

The cold transfer was roughly twice as fast as the unoptimized transfer, and the warm transfer was about 70x faster than the unoptimized transfer and 34x faster than the cold transfer.

While we were all pretty excited about these results, I don't think anyone was more excited than Twitter user @iceklube. Jamie Klube from Mililani on Oahu in Hawaii was our big winner of the $500 Apple Gift Card! 

He guessed 280 seconds for the unoptimized transfer and 8 seconds for the optimized transfer.  Of the entrants who followed all of the instructions, his guesses were the closest, with a total differential of just 14.9 seconds.

Jamie is a big user of our Steelhead appliances in his organization and for his customers, and he says that his users know immediately if they turn off the optimization.  He watched our demo videos on the blog, and when he learned of the contest, he extrapolated those results to a 50MB file, and that's how he got so close.

So congratulations to Jamie, and thanks to everyone who entered the contest.  We appreciate your efforts and support.

Keep watching the blog and the Twitter feed.  We'll have more contests soon...