There were three big announcements this week at Amazon Web Services. All three announcements are important but the first is the one I’m most excited about in that it is a fundamental innovation in how computation is sold.
The original EC2 pricing model was on-demand pricing. This is the now familiar pay-as-you-go and pay-as-you-grow pricing model that has driven much of the success of EC2. Subsequently reserved instances were introduced. In the reserved instance pricing model, customers have the option of paying an up-front charge to reserve a server. There is still no obligation to use that instance but it is guaranteed to be available if needed by the customer. Much like a server you have purchased but turned off. Its not consuming additional resources but it is available when you need. Drawing analogy from the power production world, reserved instances are best for base load. This is capacity that is needed most of the time.
On-demand instances are ideal for Peak Load. This is capacity that is needed to meet peak demand over the constant base load demand. Spot instances are a new, potentially very low cost instance type ideal for computing capacity that can be run with some time flexibility. This instance type will often allow workloads with soft deadline requirements to be run at very low cost. What makes Spot particularly interesting is the Spot instance price fluctuates with the market demand. When demand is low, the spot instance price is low. When demand is higher, the price will increase exactly as the energy spot market functions.
Also announced this week were the Virtual Private Cloud unlimited beta and CloudFront streaming support.
Elastic Cloud Compute Spot Instances: Amazon EC2 Spot Instances are a new way to purchase and consume Amazon EC2 Instances. Spot Instances allow customers to bid on unused Amazon EC2 capacity and run those instances for as long as their bid exceeds the current Spot Price. The Spot Price changes periodically based on supply and demand, and customers whose bids meet or exceed it gain access to the available Spot Instances. Spot Instances are complementary to On-Demand Instances and Reserved Instances, providing another option for obtaining compute capacity. If you have flexibility in when your applications can run, Spot Instances can significantly lower your Amazon EC2 costs. Additionally, Spot Instances can provide access to large amounts of additional capacity for applications with urgent needs. To learn more, please visit the Amazon EC2 Spot Instances detail page.
Amazon Virtual Private Cloud Unlimited Beta: Amazon Virtual Private Cloud (Amazon VPC) is a secure and seamless bridge between a company’s existing IT infrastructure and the AWS cloud. Since August 2009, Amazon VPC has been in a limited beta, during which we’ve selectively granted access. Starting today, all current and future Amazon EC2 customer accounts are enabled to use Amazon VPC, but customers will not be charged for Amazon VPC until they begin using it. Amazon VPC enables enterprises to connect their existing infrastructure to a set of isolated AWS compute resources via a Virtual Private Network (VPN) connection, and to extend their existing management capabilities such as security services, firewalls, and intrusion detection systems to include their AWS resources. To get started with the service, please visit the Amazon VPC detail page.
Amazon CloudFront Streaming: Amazon CloudFront, the easy-to-use content delivery service, now supports the ability to stream audio and video files. Traditionally, world-class streaming has been out of reach of for many customers – running streaming servers was technically complex, and customers had to negotiate long- term contracts with minimum commitments in order to have access to the global streaming infrastructure needed to give high performance.
Amazon CloudFront is designed to make streaming accessible to anyone with media content. Streaming with Amazon CloudFront is exceptionally easy: with only a few clicks on the AWS Management Console or a simple API call, you’ll be able to stream your content using a world-wide network of edge locations running Adobe’s Flash® Media Server. And, like all AWS services, Amazon CloudFront streaming requires no up-front commitments or long-term contracts. There are no additional charges for streaming with Amazon CloudFront; you simply pay normal rates for the data that you transfer using the service. Visit the Amazon CloudFront page to learn more.
b: http://blog.mvdirona.com / http://perspectives.mvdirona.com
Spreading the cost of FMS over a very client base is one of the strengths of cloud service. There are economies of scale at all levels.
James Hamilton
jrh@mvdirona.com
I knew you would like this pricing announcement Greg since you had I have talked about variants of the approach for quite some time.
James Hamilton
jrh@mvdirona.com
Mike, I agree. The posting I think you may have been referring to is Resource Consumption Shaping: (//perspectives.mvdirona.com/2008/12/17/ResourceConsumptionShaping.aspx).
It seems interesting that if you want to run FMS on your own server it costs big bucks but from a CDN it’s free. That’s a huge incentive to use a CDN.
Excellent news on the spot pricing for EC2! Very cool to offer variable pricing based on demand.
It would be interesting to discuss the algorithm for determining the spot price, since it’s probably not just maximizing revenue. You want to sell unused capacity, but maybe sometimes you want it to be unused.
If you have a load that you can shed, you can use new strategies for energy efficiency. You can run hotter because you know you can dynamically reduce load. The worst case scenarios don’t look quite as bad. You’ve said all this before, but I can’t find the specific posts.