Amongst the launch hype for Amazon’s new tablet, the Kindle Fire, was an interesting revelation about how this device will access the internet.
Mobile Internet Devices (MIDs) have to provide excellent service to their users, but are constrained to doing this using a “whatever’s available” approach to communications infrastructure. This causes problems for device architects, and Amazon has chosen an interesting route to overcome it.
A standard MID (in fact pretty much any internet device outside of an enterprise) accesses web servers through a direct connection to the internet. Everything that makes up the connection (wireless networks, 3G, broadband links etc) is transparent to the user.
Amazon provides (in their 3G enabled devices) the use of the Amazon Whispernet, which has been successfully proven by users of the Kindle eReader.
In order to address the comms issue, Amazon has decided to turn the power of its EC2 cloud to the task of being a massive proxy server. EC2 is just a collection of servers in big rooms spread across the world, and while access to this collection can be purchased by normal customers, it’s not really been a massive commercial success as cloud adoption is still in the early stages. Amazon has clearly decided to turn EC2 into a money-making machine, but in a different manner than originally intended.
Using EC2 as a proxy, the Kindle Fire accesses the internet thus:
The Kindle Fire user requests a page from a web server, but instead of fetching the page itself, the page is fetched by servers within the Amazon EC2 Private Cloud. This offers benefit to the user in that there’s considerable scope for caching content: if 10,000 Kindle Fire users access the same web page, this can be fetched from the web server just once, then repeatedly served from within the EC2 cache. Traffic is reduced, and Kindle Fire users get their data much faster.
But the servers shown in the diagram above are relatively anodyne. It’s probably not of much interest to Amazon which stories its users read on the BBC News website. Let’s add some servers in:
It’s difficult to get to any website without having a Google Ad puked at you, so there’s lots of traffic there. And MID users are getting a reputation as inveterate shoppers.
Think of this data being proxied through EC2. Just think about it for a minute…
Amazon sells a device to one of its customers, then this customer repays Amazon by replaying their entire browsing activity through an Amazon server. Advertisements, content and pricing from other online merchants are brought to Amazon by people who have proven themselves to be Amazon customers.
Everything a Kindle Fire user retrieves from the internet will pass through an Amazon system, and you can be sure that if the data is of interest to Amazon, they’ll keep a hold of it.
This isn’t a post screeching about user privacy. I genuinely don’t believe that Amazon is interested in anything other than selling people stuff, and their “Recommendations” have always served me well. This is a post pointing out that being an online merchant is changing. Amazon, by putting in a back-end infrastructure that most users won’t know anything about, can gather invaluable research data on those users, then use that to drive additional sales.
And let’s not forget that Apple is building its own iCloud infrastructure. How long before they route all iOS communication through iCloud?
The internet as a benevolent disinterested independent data-carrier is under threat of extinction.