Amazon.com Inc. is the world’s largest online retailer. After starting off as an online book seller, today it offers everything from electronics and apparel to toys and furniture.
The company earned revenue of $7.56 billion (Rs 33,566 crore) and a net profit of $231 million in the three months ended September, it said on Sunday. In 2002, Amazon started offering remote computing services for other websites called Amazon Web Services (AWS). It counts Indiagames.com, Rediff.com and 8kmile among its Indian clients. Amazon.com, in its latest earnings call, also announced one year’s free access to its most popular Elastic Cloud Compute (EC2) service to clients across the globe. In an interview, chief technology officer and vice-president Werner Vogels, who is on a visit to India, spoke about the company’s strategy for its AWS offerings. Edited excerpts:
Have Indian clients been different from those in other markets in terms of cloud computing adoption?
I don’t think the markets are so different in India. The only thing is (that) there is a very strong SMB (small and medium business) segment in India, both in traditional manufacturing and Web, which is hungry to use the enterprise network technologies not available to them so far. They are served very well by cloud computing.
Many companies in this segment were early adopters. Rediff is the most visible example. Another company is Hungama, which delivers digital streaming and has 80% of its IT (information technology) on the Amazon platform. Redbus.in runs 100% of its infrastructure on AWS. They originally ran out (of) their own data centre but they felt completely hampered by their inability to acquire their own hardware. Within weeks of using AWS, they saw a 50% growth of their business.
One of the driving factors for the adoption of the cloud model is that it cuts down on cost. Once the economy worldwide recovers, do you expect demand for cloud services to go down?
Cost is only a part of the picture. Many organizations traditionally see a utilization of 15% in one data centre, which means 85% of their compute power is wasted. With the cloud, you see an improvement of 75-85% in costs—so cost is important. But for most enterprise customers, agility is the main reason to get on the cloud. When resources are not an obstacle, the time to market is much shorter. Traditionally, the lead time for acquisition of servers or storage is weeks if not months—but the cloud gets access to the resources in a matter of minutes.
Increasing competition, compressed timelines, increasing consumer power, and reducing cost of production and distribution are things that have changed forever. They are a longer economic pattern which won’t change even after recovery. The only way to deal with it is by going to a completely different resource model where you can focus on the core competency. Corporations spend 70% of their mindshares and best brains to run their infrastructure today. This cannot be done any more.
Also, in terms of delivering products in different geographies, with AWS, you can simply move an application to a different market and location without having to acquire network bandwidth, computer power etc. Also, with a recovery, as available capital to customers increases and innovation picks up again, more companies will move to the cloud.
You recently launched the free-tier offering to clients worldwide where they can use the EC2 service free for a year. What are the economics of this for Amazon?
We want to remove as many barriers as possible for people entering the cloud. There is a whole range of smaller customers who want to try out the cloud. A student implementing a great idea who doesn’t now need to make a decision between paying for his server and paying for his food. It could be an engineer who always wanted to try a new version of a content management system. We also want to bust myths like the cloud not being secure, reliable, or that when Christmas comes, Amazon will take all the bandwidth away from you.
As far as capacity utilization goes, Amazon has a very innovative payment schedule. We don’t do just basic utility pricing—such as on-demand, pay per hour. Customers who run 24X7 (and) 365 days a year can use something called reserve instance pricing. If they commit to their usage, we drop the pricing to 40-50% of the original cost. In spot instances pricing, customers can set the price they are willing to pay. If it drops below what the customer sets, we take on the jobs for them. So in our model, we have a layer of reserve capacity. On-demand sits on top of that, and whatever is left, we fill up with customers from our spot market. So, no matter how much business is generated by the free tier, we will always have perfect utilization.