Last May, at the Content Delivery Summit event I organize each year, Amazon Web Services (AWS) announced support for dynamic content delivery, also referred to in the industry as dynamic site acceleration (DSA) (video presentation here). Since the launch, Amazon (AMZN) has been pretty quiet on what they have been working on, so I spent some time with the company to get an update on where their support for dynamic content stands. For those not familiar with DSA technology in general, dynamic site acceleration is a suite of technologies and products that deals with optimizing dynamically served content across a network. Traditional DSA services often include TCP optimization, route optimization, connection management, on-the-fly compression, SSL offload and pre-fetching functionality amongst other technologies.
Amazon is targeting customers with their dynamic content delivery service for two distinct applications: content that changes frequently -- e.g., sports scores, weather updates, stock quotes, etc. -- but may be cached for short periods of time, and content that represents the interactive or personalized portion of a website, created on-the-fly for each user and generally cannot be cached. To coincide with their new dynamic content delivery product, Amazon's CDN CloudFront launched several new features to support their DSA offering.
These features include configurable minimum time-to-live (TTL), query string caching, support for multiple origins, and URL based cache behaviors. With configurable minimum TTLs, dynamic content can either be cached at the edge for very short periods of time or not cached at all. With query string caching, customers can cache dynamic web pages, such as search results pages, product pages, etc., at the edge for any amount of time. With support for multiple origins, customers can point at different origins for different types of content on a single website -- e.g., Amazon S3 for static objects, Amazon EC2 for dynamic content, and a custom origin for third-party content. In addition to these new features, all previous Amazon CloudFront features (e.g., private content, invalidations, default root object, etc.) are also available for Amazon CloudFront accelerated dynamic websites.
Since May, Amazon's been busy improving their dynamic content offering and CloudFront has added a number of new features to enhance the dynamic delivery. These features include:
- Cookie support. Just like query strings, cookies are also used by customers to return dynamically generated content. With Amazon CloudFront, customers can now configure cookies to be part of the cache key so that the origin can personalize web pages for each viewer based on specific actions – e.g., response to a web form or after the viewer has logged in (see here).
- Front-End-Optimization. AWS partnered with Strangeloop Networks to offer front-end-optimization (FEO) for dynamic websites. A joint customer already using this solution is Outrigger.com, as referenced in the AWS blog.
- Lower Pricing. On Feb. 1, 2013, AWS lowered the price of data transfer from AWS Regions to Amazon CloudFront edge locations for "origin fetches," in some cases by as much as 83%. This includes data transferred from Amazon EC2 and Amazon S3 to any Amazon CloudFront edge location.
- New edge locations. Since May 2012, AWS added nine new Amazon CloudFront edge locations, including new markets -- Madrid, Spain and Sydney, Australia -- to help improve performance for all Amazon CloudFront customers around the globe.
While Amazon has been busy improving their dynamic content offering, they still have a long way to go before it has a lot of the functionality Akamai's DSA offering has, including support for commerce customers. But as with all of Amazon's Web Services, it usually takes the company 18-24 months before their service competes for a large percentage of the market. One of the biggest reasons companies have taken note of Amazon's new dynamic content delivery service is the fact that Amazon doesn't charge a premium for their service, something most other vendors do. Amazon's dynamic content pricing is the exact same price as their CloudFront CDN pricing. There are also no platform fees, setup fees, overage charges, and many customers find that they don't need to use expensive professional services to configure dynamic applications, further decreasing the cost of using CloudFront for dynamic content delivery.
Since launching eight months ago, Amazon says thousands of customers have adopted Amazon CloudFront's dynamic content delivery features. The company won't discuss the breakdown of their customers, but I know that for now, they are mostly small customers who don’t pay a lot each month. But these customers represent a variety of use cases, including news portals, sporting websites, local/weather applications, targeted advertising systems, as well as social and gaming applications. For many of these customers, accelerating their entire website (including the dynamic portions of their site) was cost prohibitive in the past because they weren't large enough to use a service from someone like Akamai. But with its cost-effective pricing model for dynamic content delivery, Amazon CloudFront is helping democratize this market for small customers and it's only a matter of time before they start targeting and attracting much larger customers.
Just as AWS has done in the past, they plan to continue to rapidly add new functionality to their existing dynamic content delivery capabilities based on customer feedback and further additions to the service will be rolled out later this year. This includes performance improvements as well as adding new features and more edge locations around the globe.
Many have been quick to imply that Amazon's dynamic content delivery service is too bare-bones to be able to compete for business from large media and enterprise customers looking for DSA services, but that’s a dangerous assumption to make. When Amazon's CDN product CloudFront launched, it was in a beta stage for about 18 months with little functionality and many suggested it would never compete for the large RFPs in the market. But 18 months after launch, Amazon rolled out tons of additional product functionality to CloudFront, added in an SLA, lowered pricing multiple times and by my estimates (as well as their competitors), did more than $100 million in revenue last year. So while Amazon still has some ways to go with the product, they've entered the market quickly and have the skills and resources to compete for the long-term.
Amazon's dynamic content service will be no different. Sure, it's missing a lot of the functionality that Akamai and others offer today, but give it time. Amazon is in no hurry and can roll out more features and functionality in a steady and systematic way. Amazon's not trying to dominate the market in the short term, they are setting up the foundation to compete for a large portion of the market over time and they have all the resources to be able to accomplish it. So while many may write them off today, they won't in the future.
What impact Amazon could have on DSA pricing in the market and how that could affect someone like Akamai, who dominates the industry with DSA and web optimization services, is still unknown. I'll have more thoughts on that topic in another post. But one thing I would not do is bet against Amazon or not take them seriously as that would be a mistake. They have the resources and skills to impact any market they enter, including the web acceleration industry.