AWS Summit NYC, Day 2, keynote

In the morning of day 2, we were led into the very large room for the keynote. There were probably 1000-1500 people in the hall, sitting in tight packed formation. Amazon VP & CTO Werner Vogels entered the stage with fanfare (the hall darkened, upbeat music blasting from the speakers). Werner gave a superb, clear, and engaging talk. You can read my notes of it below, or watch the keynote here (around 2 hours long).


The AWS summit series kickstarted in NYC, because there is a good tech crowd here, and more importantly many AWS customers are in NYC. AWS summit series this year will be in 13 cities, for which a total of 20K people registered.

It is hard to believe that AWS S3 launched only in 2006. There has been amazing progress in 7 years in AWS. Now there are a total of 33 services running, which span compute, storage, database and application management. 100,000 businesses are using AWS This list includes even an extraterrestrial client: NASA JPL is a good customer, and Curiosity rover data flows to AWS, where many customers access this data for data-mining.

AWS provides a wide-range of technologies to avoid lock-in to a specific OS and database, tool, and programming language. Consulting partners and technology partners are growing at a tremendous rate. Last year AWS had launched the AWS marketplace, where you can search for software that is prepackaged to run on AWS. Today there are 25 categories, 778 product listings on the AWS marketplace.

AWS takes pride in their focus on customers; they had 31 price reductions since 2006. Economies of scale (and also Moore's law) draw down the costs, and AWS tries to return these back to the customers (this is actually an investment to grow faster). Werner is visibly excited by this: "Wider adoption due to lower prices will enable tremendous beautiful applications, since you will no longer be constrained by the infrasture or usage costs". Werner believes that "the real growth is just around the corner, this is just the beginning".

AWS services

Next Werner goes on to say a couple sentence for some major AWS services.

S3 (simple storage service) recently crossed over 2 trillion objects. It has seen quadratic growth since 2006, and is routinely serving more than a million storage requests a second.

DynamoDB provides managed non relational database that is consistent, high performant, and scalable. AWS focused on the database because traditionally the database has always been the bottleneck in scaling. DynamoDB provides auto-scaling, with low-latency. It is indexed by key and also supports ranges on the key. In the keynote, local secondary index support for DynamoDB is announced. Werner said "Now that, the database is no longer the bottleneck, you can focus on your application to make it scalable".


Next Werner started to talk about the transformation AWS enables: what drives it, who is doing it, and what is next?
The most radical and transformative of inventions are those that empower others to unleash their creativity and pursue their dreams. -- Jeff Bezos 

What drives this Transformation

The success of cloud is because it fell into an economic model where there is an abundance of products, intensifying competition, significant reduced consumer loyalty against products, limited capital. These drive tremendous uncertainty about whether your product is going to be successful. In times of uncertainty you need very different resource models; you need flexibility to address these uncertainties: acquire resources on demand, release resources when no loger needed, pay for what you use, leverage other's core competencies, turn fixed costs into variable. IT wanted to do this for a long time, AWS enabled it. AWS provided the following benefits.

  • AWS help trade capital expense for variable expense to get started: E.g.,Samsung TVs run software that needs to connect to backend infrastructure, so Samsung went with AWS, saved millions of dollars in fixed capital expense.
  • AWS provides lower variable expense than companies can do for themselves.
  • You no longer have to guess capacity.
  • Dramatically improved speed and agility. If you want to increase innovation you have to reduce the cost of failure in AWS you can do many experiments at the same time.
  • Stop spending money on undifferentiated heavy lifting.
  • Go global in minutes.


Who's doing it

Every industry! AWS help achieve radical transformations in every industry.

In insurance business, the climate corporation is building completely radical approach to insuring farmers. They do weather insurance, but no agent is sent to physically visit the farms. This is done in a completely data driven manner. Once a month they run 5K cores on AWS Elastic MapReduce, go through 10K scenario models, spinning 20TB of data. Climate coorp couldn't do this on their premise, massive amount of capital needed to build this infrastructure themselves.

Financial services uses AWS (see the customer talks below for Nasdaq's talk). This is a challenging industry, where there is extreme competition, and stringent regulatory compliance/security requirements.

Media and advertising companies use AWS. Werner said "almost any media you think is AWS customer" and counted among others Fox, PBS, ABC, Newsweek, Washington Post, Independent, Reuters. ABC uses AWS to stream media for exactly the device you are watching on (iphone,ipad, computer) and perform context/location-aware ad insertion. The poster child is of course Netflix, the media streaming giant which has all their operations on AWS. Netflix provides their cloud tools as open source, and sponsor a Netflix open source prize which is in progress.

Manufacturing which has become increasingly global and highly collaborative requires significant scale. AWS powered Boeing for designs. Autodesk another customer. General Electric another customer gave a brief talk, found at the bottom of this post.

Healthcare and biotech has been seeing rapid innovation and has global scale collaboration needs. Center for disease control (CDC) collects data about health risks, does realtime flu tracking and shares with healthcare workers through AWS. Unilever is using AWS to study armpit bacteria to be able to build better deodorants, they also study oral bacteria for producing more effective toothpaste. AWS S3 hosts and makes publicly available 250TB of data from the 1000 genomes project to help research. Illumina, a biotech company, uses AWS in their efforts on sequencing of human genome. Illumina produces DNA sequencing instruments, and using AWS they provide a service where their devices publish data to cloud application. This has uses in cancer treatment: everybody's cancer is different, treatment should be individualized. Illumina's customers are biologists, who using Illumina's basespace portal can now share data with ease rather than resorting to shipping harddisks.

Hospitality sector (hotel IT) is also seeing rapid growth and opportunities. Four seasons hotel, Intercontinental, and Hotelogix are AWS customers. Airbnb, hosted on AWS, upsets the hotels, and enables you to book room/apartment directly from the owners.



What is next?

Werner starts by talking about big data trends: the move to realtime information and deeper integration. These lead to internet of things. Connected devices are incredible data generators.

Shell oil company, another AWS customer, is experimenting with new sensors that generate 1TB of data each, and they are planning on dropping 10K sensors around the world. (Seems like fat sensors is becoming the new trend.) The Mount Sinai hospital uses sensors in hospital for management (beds, doctors, patients, etc).

Your music/books/content is now in the cloud, and devices serve just as a window to that.  Werner wants the treadmill machine in the gym to be like that, another window to the cloud.  The treadmill should realize who I am and set up automatically.

Smartphones are also a primary big data generator and first class citizen in internet of things. Smartphone location-aware services are very significant, and the development of these apps will be dramatically transformed by cloud support. The apps will just stitch a web of services together, this will be the modern style of app dev.

As a final point, Werner mentions security. AWS has readily available encryption services to protect customers and they want to see that encryption is dead simple to use. For example on amazon.com web site all traffic and storage is encrypted. He announce AWS Relational Database Service (RDS) support for transparent data encryption for Oracle.


AWS customers talks

To support his points Werner invited, spread around in his talk, 4 AWS customers to talk about their use of AWS. I am summarizing each of these customers' uses-cases together below.

Bristol-Myers Squib: They service to clinical pharma, molecular dynamics, and computational genomics. They use AWS to support scientists running clinical trial simulations, which are very computationally intensive. Using AWS provides them to have on demand compute capacity, time and money savings, agile innovation, and reduced patient burden (fewer subjects, fewer blood tests).

Nasdaq: Nasdaq is synanomous with stock exchange, but nasdaq also provides IT solutions to industry as their business. Using AWS, they launched FinQloud platform which provides financial-services customers with a secure financial-industry only cloud.

General Electric: They talked about a platform they will build: cloud enabled crowd-driven ecosystem for evolutionary design (CEED). There were no details or concrete applications, but CEED will help address the design and creation of complex systems that require global collaboration. (Yes, very vague I know.)

Mortar: Mortar is a big data startup that won the AWS global startup challenge competition this year.  Mortar, founded on 2011 on AWS,  provides "Hadoop as a service". They aim to democratize data analytics leveraging on AWS which democratized infrastructure.

Comments

Amee said…
This is cool!

Popular posts from this blog

The end of a myth: Distributed transactions can scale

Hints for Distributed Systems Design

Foundational distributed systems papers

Learning about distributed systems: where to start?

Metastable failures in the wild

Scalable OLTP in the Cloud: What’s the BIG DEAL?

The demise of coding is greatly exaggerated

SIGMOD panel: Future of Database System Architectures

Dude, where's my Emacs?

There is plenty of room at the bottom