Tis the week of Christmas and all through the house, not a creature is stirring except to click their mouse. Black Friday and Cyber Monday are far behind us, but there are still a few doing their last-last-minute online shopping. This entire week, the tubes were getting clogged due to this. Akamai’s Retail Index shows that shopping traffic peaked on December 1st at around 8 million simultaneous shoppers. Retail sites of course are perfect candidates for content delivery networks, who have the scale to absorb the holiday bandwidth spikes.
Here’s a diagram showing the traffic patterns:
When the tubes get clogged, those who make their living selling things on the net can be in real trouble. Several studies have shown a strong correlation between conversion (moving someone from merely browsing to buying) with site responsiveness. Forrester Research recently found that consumers’ expectations for page load time have dropped in half over the last few years, with half of consumers being willing to wait no longer than 2 seconds for a page. Poor performance can have permanent market-share effects: 79% of consumers who had a bad experience on a site are likely to no longer shop there.
Google’s upcoming algorithm changes mean that any retailer with performance problems on its site could be affected, as this will now be a key part of the the mix that determines their all-important Google page ranking. See this interview with Matt Cutts of Google for more information on this.
There are several ways to improve site performance: CDNs move caches closer to users, and good code design can circumvent some limitations of HTTP. If we peer into that traffic, we find that only about a quarter of the traffic is text information, and most of the bandwidth is consumed by images. Therein lies another opportunity to reduce page load times, by optimizing images.
We at Ocarina have been getting to know some of these retailers of late. As you may know, our main focus is storage cost reduction and improving storage utilization. However, we began to run across customers with relatively little storage – at least not enough to constitute a cost or scalability pain who nevertheless were excited about what we had to offer. These were businesses that recognized that our technology had the immense benefit of improved end-user experience.
The typical Internet retailer may have tens of thousands of images (or millions in some cases) in 10-30 TB of storage. These customers use Ocarina’s Native Format Optimization (NFO) workflow, where files are shrunken in their native format, and retain all the savings completely through distribution to the end-user. If we shrink their images by 45% on average, that can literally mean page load times 1/3 faster, which is a huge benefit, not to mention the substantial savings in bandwidth. A happy holiday tale indeed.
