Amazon CloudFront: No Threat to the Incumbents
Amazon announced on Tuesday a new S3-based service called CloudFront. This content delivery network that aligns with the philosophy of Amazon's cloud services: easy to get started, incremental measurement of use, and pay-by-the-drink pricing.
You put a file (i.e., content of some sort, for example, an image file) into S3, create a distribution that, you guessed it, distributes copies of that file from the originating S3 location, and then put links to the object in your web application using the distribution name. Amazon then arranges to move copies to its edge locations across the United States, Europe and Asia.
A lot of discussion about the service has, as you might expect, gone on in the blogosphere. One interesting posting by Dan Rayburn, Principal Analyst at Frost and Sullivan, is titled "Why Amazon's CDN Offering is No Threat to Akamai, Limelight or CDN Pricing." It's a pretty interesting piece and argues against those who, apparently, believe CloudFront will decimate the major CDN networks. Rayburn points out some reasons why CloudFront is no threat to the incumbents:
1. It doesn't have the reach they do: They have zillions of machines scattered around the globe located inside ISPs. By contrast Amazon has an unknown number of machines in a limited number of locations, all of which are Amazon data centers. Rayburn doesn't say this, but this means content retrieval requires an additional hop from the end users ISP to the Amazon location.
2. The incumbents aren't interested in the customers Amazon is targeting. The incumbents want customers who are ready to spend hundreds or thousands of dollars per month. It's not economically feasible for them to deal with a customer who might spend $100 or so per month.
3. At really large scale, Amazon's pricing is actually more expensive than the incumbents, who can offer big discounts for high usage.
Rayburn then points out that markets aren't homogenous; both Hyundai and BMW sell cars, but really don't compete.
He's right to dismiss the postings that posit CloudFront as a threat to Akamai or Limelight. Someone with huge content delivery needs that come from all over the world would find CloudFront inadequate, perhaps laughably so. However, I think he sells the impact of CloudFront short, relegating it to a Hyundai role (leaving aside the fact that many people purchase BMWs to send a message about their prosperity or self-regard, which is a less powerful motive in most business settings-not that many companies are going to be willing to spend significantly more just to tell people that they use Akamai).
I think a better auto analogy would be to cast CloudFront as a Ford-the Ford of 1913. Prior to Ford, cars were excellent ways of achieving personal travel autonomy and speed; the alternatives were travelling by horse (slow) or train (fast, but limited in places it would take you). Cars offered a huge advantage to the existing modes of transport; this is analogous to the incumbent CDNs: they're far better than the alternative, which is putting all your content on your own server (slow) or building out your own network to host content (fast, but limited in where you can place them, because, lets face it, you're not going to put them in as many places as Akamai).
However, most people couldn't afford an auto in 1913. Cars were luxury goods, available only to those with lots of financial resources. Ford designed an auto that could be assembled easily, and then implemented a new manufacturing system to drive assembly efficiency. He turned owning an automobile from a rarity to a commonplace, and in the process, changed our society forever.
Looking at CloudFront with this analogy in mind, we see similar parallels. CDNs are great, but they can only be used by people with deep pockets (indeed, as Rayburn points out, they don't want to talk to anyone lacking deep pockets). They're not very easy to deal with-you're invited to fill out a form on the Akamai site so a sales person can call you back. By contrast, CloudFront is for people who don't have a lot of money, or at least don't want to make a big financial commitment up front. It's easy to get going with CloudFront: all it takes is a few commands and a credit card.
Ford didn't destroy the luxury auto market, but he expanded the overall auto market a thousand-fold, or maybe even a million-fold. However, he did put a lot of auto companies out of business. When he started, there were several hundred auto manufacturers all appealing to the well-heeled, companies with products like the Jordan Playboy or the Stutz Bearcat. They couldn't compete at the low-end and didn't have enough of the small high-end to survive.
So I doubt that Amazon will put Akamai or Limelight out of business. They provide a great service for a demanding customer base, which has problems CloudFront is unlikely to solve. However, the players lumped into the "CDN" category-if I were them I'd be worried. Amazon is going to vastly expand the CDN market. If you're a CDN company that can't compete at the new pricing floor and can't own enough of the high-end market to prosper, well, things won't be so good.
There's a good chance we'll look back in a few years and see CloudFront as the Model T of content delivery: simple, reliable, and utterly transformative.
Bernard Golden is CEO of consulting firm HyperStratus, which specializes in virtualization, cloud computing and related issues. He is also the author of "Virtualization for Dummies," the best-selling book on virtualization to date.