This is the third entry in our exploratory series "Will One Company Dominate
the Cloud". Today we're blinking twice after reviewing the innovation engine at Amazon.
The Amazon AWS product is all about services. While others are marketing the cloud with an
explanation point, the cloud leader is focused on the raw building blocks. This includes
everything from storage to people. Amazon is learning how to find new ways to optimize
connections and monetize them in increments of time.
Sponsor
Amazon, the Verb: Motion
When thinking of Amazon as a verb, one word stands out, motion. When Amazon was first introduced
as the Internet bookstore, it immediately created a change in the landscape.
It seemed like the writing was on the wall for brick and mortar retail, and to a large degree, it
was. In a mere 15 years, it has disrupted the entire book vertical with an end-to-end digital
system. Amazon is now in the position to completely automate the flow of content bits from
upstream to downstream.
Now let's look at the AWS services to see if can it do the same for computing.
We'll analyze the services Amazon offers and how they work together, specifically in four areas:
computing, storage, networking, and people.
(Although we didn't include several areas in this roundup, including database and monitoring, we
see them as clear signs of momentum and scope of Amazon's evolution.)
Compute
We signed up (again, as a new user,) for EC2 to refresh ourselves
with its offerings and to remind ourselves what it means to be utility-based.
Amazon defines workload in relationship to the types of instances the company offers in the EC2
solution.
Windows on EC2 is
optimized around bringing a three-tier Windows web environment into the Amazon stack. It supports
ASP.Net, AJAX, IIS, and SQL Server. Amazon has also tuned it's network and storage offerings to
nicely plug into the Windows on EC2 package and offer seamless integration with existing Amazon
EC2 features like Amazon Elastic Block Store (EBS), Amazon CloudWatch, Elastic-Load Balancing,
and Elastic IPs.
IBM WebSphere is also supported on
EC2, and hosts a lineup of enterprise computing tools including the WebSphere Server, Portal
Server, DB2, Tivoli Monitoring, and Data Quality products. IBM mentions that one of the targets
is getting developers to use this model for getting development or proof-of-concepts projects up
and running quickly.
The patterns for firing up a new instance are defined as AMI (Amazon Managed Instances) so the
software has been appropriately targeted the infrastructure instance it will run within. IBM and
Amazon have set up Have extra licenses, or want to retire legacy hardware? IBM has an agreement
with Amazon to allow you to migrate your licenses to EC2.
The EC2 MapReduce is a service that targets large data
streams and optimizing processing of these data sets. It leverages the Hadoop Map Reduce project and provides as an example of
breaking the computer entirely into services.
The Map Reduce service doesn't just host an application stack, but is automatically configured
using Amazon Simple Storage Service (Amazon S3). This is an example of an open-source
implementation project (though Apache) optimizing in such a way that it fits on the EC2 stack as
a core feature, and it has become a peer to the WebSphere or .Net patterns.
Storage
The storage offerings include S3, Elastic Block Storage, and Input/Output.
Amazon S3 (Simple Storage Service) has been out there
several years serving web based applications as their simple cloud away from home. Customers of
it have famously stood up their entire data solution for images and other key storage tasks based
on Amazon's S3 service. It's popular, well known, and evolving to include additional features
that enforce data level integrity like databases.
Elastic Block Storage is another storage service offered
by Amazon. Instead of being a simple, writable data service in the cloud like S3, it is focused
on EC2 instances that need storage as part of their footprint. An EBS can be built alongside the
EC2 instance that is 1GB to 1TB in size and can be mounted from that service. This is designed
for applications that expect raw physical storage locally addressed by the server.
Network
Amazon offers Elastic Load Balancing. Considering Amazon's
power as an elastic compute provider, this is a critical piece of the puzzle. Here, load can be
configured to continually monitor and self heal across a set of hosts, moving the resources
towards optimal performance.
The company also offers Virtual Private Cloud, which
enables an enterprise to segment access to a portion of Amazon's cloud with access control and
security enforcements (such as subnet, encrypted VPN).
People
An amazing thing about all of these services coming from Amazon, is that Amazon is a consumer
facing company with an amazing relationship with consumers.
Amazon has the ability to learn about us. We share our ideology (books we buy), lifestyle
(products we consume), and financial position (credit cards we use). The company has also
implemented an important part of identifying consumers by going deeper with services and
verifying identity.
The company implements a two-factor signup process that goes the extra step in granting
authorization to a user to change compute resources.
This second factor gives Amazon some assurance that the person really is that person, because in
addition to getting the credit card and password (which are network resources), it also calls out
to your phone to verify that the person logging in to the network has the phone (physical
resource) at the same time.
Here is step one: Signup
Here is step two: Verify PIN on your mobile phone:
And, step three, proceed (you are now free to spin up resources):
When combining these two things together, Amazon is in a position to easily bring its current
customer base to a two-factor security solution, and providing a service that meets government
level controls. And, with two factor credentials it's less likely that there will be automated
bots being deployed in Amazon's cloud by scripts or hackers.
Amazon is in the unique position to view the next generation computing fabric from the consumer
sales process. Amazon may be the only company in a position to see how it all pieces together,
even perhaps a longer view of the future supply chain than its new book competitor, Apple.
In addition to consumers and developers, Amazon also has the power of people as resources, with
the Mechanical Turk marketplace.
Need a simple task completed and queued for the Internet (of people) to execute on? Get started
with one of these sample scripts and draw legions to your command.
We find it compelling that Amazon has connected consumers, verified individuals, and tasks to be
executed on. These pieces are perhaps foundations for a broad appetite for connecting workers
with resources and optimizing along with way.
Banking with Amazon - or - Selling Time Instead of Licenses
The time value of money is the
value of money figuring in a given amount of interest earned over a given amount of time.
When signing up for the AWS features as a new user, we found ourselves asking looking at pricing
options that reminded us of bank products. Earn more by committing to 1, 2, or 3 years. Are the
Amazon Web Services an economy, and the individual services themselves currency?
First, let's look at Microsoft and its revenue. A server is sold, Microsoft gets a piece by the
sell of the OS. Part of this business model is very predictable (company gets x% of all PC
shipments. And part of it is a bit lumpy. Where consumers have choices, they may choose to
exercise them. For example, choosing Google Docs as an alternate to Microsoft Office, or
bypassing an entire OS update, such as Vista. These choices represent risk to Microsoft in its
revenue position.
Amazon, is increasingly using something more predictable to sell it's services, time. And the
nice thing about time, is that it's always ticking. So, instead of waiting for an entire "new
PC", or "OS update", Amazon's implementation of selling resources is triggered to contracts. And,
if this works, the consumer of the risk chooses the service longevity and the risk is reduced for
Amazon.
To put this in financial terms, the time value of money states. "The method also allows the
valuation of a likely stream of income in the future, in such a way that the annual incomes are
discounted and then added together, thus providing a lump-sum "present value" of the entire
income stream."
What this means, is that Amazon is going to understand value for its AWS users over the entire
life of their contract and can start to model interaction patterns against future events. For
example, if Amazon knows you have a 3 year contract for EC2, but you're 50% more likely to renew
it if it also has SimpleDB services, it can trigger events and discounts based on these service
connections. Here we see the EC2 reserved instance pricing chart. There is heavy discounting for
committing to a term.
From what we see, Amazon will be successful in gaining new efficiencies in pricing of computing
resources, like it did with books. We expect the company to successfuly squeeze out hard costs
that exist in the middle.
We feel that Amazon is the quiet cloud company that you can "go long" with in terms of it's
future value. Like the market itself, Amazon is a prime innovator in sharing the future into the
terms of the present.
Will cloud computing re-factor how we look at the technology stack for good, and will "payment"
be in the middle? If so, is time the business model?
Photo credit: wwworks
Discuss
