Report

AWS re:Invent: recap and analysis

Table of Contents

  1. Summary

Summary

Amazon’s first-ever cloud-computing conference, re: Invent, sought to convince CIOs that the company’s cloud is now robust enough to produce enterprise-level workloads from a reliability, security, and performance standpoint.

A walk through the expo hall of the Venetian Las Vegas revealed just who makes up Amazon’s current customer base: a mix of developers, partners, vendors, and would-be cloud users, totaling close to 6,000 in number. The place was packed with 25- to 35-year-olds feasting on beer, steak in a cup, and cheesecake lollipops.

Above all, re: Invent felt like a microcosm of the shifting power in the current IT market. The old-guard technology giants had the smallest booths at the conference, while smaller cloud-computing upstarts occupied the largest. And everyone seemed incredibly charged and excited to be in on the disruption.

These startup entrepreneurs and early adopters of Amazon Web Services (AWS) generated $1.5 billion in revenue for Amazon in 2012, according to Wall Street analyst estimates. That’s an awesome feat for a company that started out as an online book retailer. But Amazon’s cloud revenue, if accurate, is peanuts compared with global enterprise IT spending overall, which is expected to hit $2.6 trillion in 2013, according to Gartner.

Eyes are on the enterprise

Amazon clearly wants a piece of the enterprise IT pie. That’s why it spent the majority of its conference showcasing its enterprise capabilities. It leaned on the big guns already using its service — NASDAQ, NASA, the U.S. government, and Netflix, among others — to help sell its story to other enterprises.

AWS Senior VP Andy Jassy showed a slide, noting AWS’ integration with VMware images and Microsoft Active Directory, the dominant service used in the enterprise to authenticate users. He listed a handful of compliance certifications that Amazon has under its belt as well as its Direct Connect and Virtual Private Cloud services, which are aimed at enterprises looking for hybrid cloud connectivity.

But Amazon’s cloud services are still a long way from enterprise-grade in the minds of many would-be users.

Even Netflix CEO Reed Hastings, the poster child for AWS, said live onstage that the service wasn’t simple enough to use. “We are still in the very primitive first assembly-language phase of cloud computing,” he said. “When you have to pick M1, M2 instances, you know something is wrong.” He was referring to the virtual-machine sizes users buy on Amazon’s Elastic Compute Cloud (EC2). He added that live migration of EC2 instances for failover between regions in the event of a disaster was also important, though currently missing from EC2. He said this was something VMware had been able to do in the enterprise for a while but that’s hard to do at scale.

In short, there’s no doubt that Amazon has its work cut out to reach the level of sophistication in workflows, orchestration, policy management, tooling, compliance, governance, and security that has been crucial and commonplace in enterprise IT shops for decades.

Low cost is not the be-all, end-all

The story Amazon most often tells is about the low cost of its cloud service. Yet for many large enterprises with big IT budgets, price is often not the most important factor. Amazon founder and CEO Jeff Bezos is right when he said during his keynote that none of his customers ever tell him they want prices to increase. But that’s because his customers are buying Kindles and other consumer products, where price sensitivity is paramount. Enterprise CIOs have other priorities, and they are often willing to pay more for the guarantees they rely on from traditional enterprise IT software and services.

There is another aspect of the pricing story that might trouble Amazon down the road. AWS faces a war with Google, which dropped the cost of its cloud storage by 20 percent  the day re: Invent started. AWS countered with a price drop of its own the same day. This battle is a concern because it’s not clear that AWS is turning a profit yet. According to Baird analyst Colin Sebastian, AWS requires significant capital expenditures — roughly $500 million, or about 50 percent of its total 2010 capex, he wrote in a note to investors. With deeper pockets and a larger research and development budget Google could turn up the heat on Amazon at any time.

Ecosystem tension as AWS launches competing products

Still, the retail giant appeared unabated. In addition to price cuts it launched a new product at the show called Redshift, a petabyte-scale data warehouse in the cloud. The service costs one-tenth the price of a traditional on-premise data warehouse appliances, at $1,000 per terabyte per year. It’s powered by ParAccel, a company AWS invested in back in July 2012.

The Redshift launch has some potentially worrying consequences for smaller cloud-based data-warehouse companies running on AWS, such as Treasure Data and BitYota. Customers, after all, are less likely to trust a small, unknown startup and more likely to go with Amazon’s service. Redshift also muddies the waters a bit for Amazon’s own Elastic MapReduce (EMR) big data analytics service. Why fiddle about spinning Hadoop clusters when you can use your existing BI tools on top of Redshift? Amazon claims 4 million Hadoop clusters were launched on EMR in the past two years. Will that adoption slow down? My suspicion is that enterprises will try out Redshift while academic institutions and the open-source community, which are already comfortable with Hadoop, will continue using EMR. Certainly no one is complaining about all the innovative new options available.

Amazon CTO Werner Vogels touted a new workflow orchestration service in the making called Data Pipeline that will coordinate the movement of data between repositories including Amazon’s S3 storage service and the Redshift data warehouse. The service sounded fairly rudimentary compared with traditional BI workflow management and integration tools, but it’s early days still.

AWS IPO on the horizon?

To maintain its lead in 2013 and beyond AWS will need to continue innovating with new products and features as fast as it has previously. This will get harder as it moves into the enterprise, where customers prefer stability and security over the latest and greatest speeds and feeds. Furthermore, next year Google, Microsoft, Rackspace, and others will be more aggressively nipping at Amazon’s heels and will soon become viable alternatives. As Amazon begins to feel pricing pressure from these competitors it will need to find a way to realize the value it has created in AWS.

It is possible Amazon will sell a percentage of AWS via an IPO and in doing so unlock the value of the cloud business on the public market. Generally speaking, spin-offs are good news, as the sum of the parts is often greater than the whole. Spin-offs also tend to liberate hidden value. This is how EMC successfully realized the value of VMware in 2007.

An IPO of AWS would potentially achieve several things:

  • Improve visibility into AWS’ performance and growth relative to the rest of the market (Right now investors interested in AWS have to dig through Amazon’s financial documents for clues into the performance of AWS, as the company does not break out these numbers)
  • Strengthen AWS’ employee retention and recruitment through an equity award pool
  • Grow the partner ecosystem independently of Amazon’s retail business

With the agility and freedom of a startup, AWS would be able to continue innovating at a fast pace even in the face of stiff competition from tech giants such as Google and Microsoft. The upcoming year will be an important gauge of its ability to attract enterprise customers and could also be part of a larger plan to prove its longevity and viability as a stand-alone business.

Access Report

Available to GigaOm Research Subscribers

Subscribe to
GigaOm Research