Today, we’re announcing expansion of our network to 33 edge nodes worldwide, with support for a new node in São Paulo, Brazil. We now have nodes in more than 16 countries worldwide, including recent additions in Dubai, Madrid, Paris, Johannesburg, Montreal, and Toronto.
In testing, we’ve observed a 3x reduction in time-to-first-byte and as much as 5x increased throughput for many Brazilian markets with the São Paulo node. Of course, real world results will depend on user connectivity, site-specific design decisions and other localized factors. But it should lead to increased performance for many end users.
Building a global product is something that is very important to us. In fact, it is a big part of what led me to start this company in the first place.
When I was at YouTube, I built an ultra-lightweight version of our Watch page, code-named Feather. In a world where someone built a clone of the video game Quake that had a file size of 100KB, the thinking went, we should be able to play a video on a page that was at least as small. We succeeded. But what we had not expected was the effect it had on user behavior. Shockingly, the average time to load actually went up, even though we had shrunk the file size and the number of requests by an order of magnitude.
A look at the analytics quickly showed us why. There was a huge influx of visitors from places we were not, at the time, accustomed to seeing in our traffic logs — places like Southeast Asia, South America, and sub-Saharan Africa.
These visitors did not have the kind of Internet speeds we take for granted in the U.S. Their average load time was over 2 minutes, even for a page that was less than 100KB in size, and this was dragging our average up. But they were watching YouTube videos — something they could not do before when a page load might have taken 20 minutes.
I think about this a lot when people ask me about global media delivery. One takeaway from that experience is that page weight matters, and indeed, that is something I have written about in the past. But there is another equally important idea: online, your potential global reach is larger than you think it is.
No matter what you are making online, it can have a global audience. That is the promise of the Internet, that information can move freely across cultural and national borders. To the extent that reality often does not live up to this promise, I think, it is because media delivery is still a difficult problem.
And that is what imgix is for. We are committed to solving the difficult problems of delivery for you, to making your visual media as accessible as possible, so you can focus on creating it. Adding new edge nodes is an important part of meeting that commitment. I am thrilled to boost our capabilities in Brazil, and am looking forward to many more additions like this as we continue to build a truly global graphics card for the Internet.