Comparing Google Cloud Services vs. Amazon Web Services at the Infrastructure Level

Google I/O is one of the few times you know Urs Hoelzle is going to speak.  Urs is the MAN behind Google's Infrastructure.  Urs is Google employee #8.  He is wicked smart, got plenty of money to do what ever he wants, and has the backing of the Google executive staff to build for the future. 

When most hear the words of "Cloud Services" they think of Amazon Web Services.  Amazon.com being a great retailer has a phenomenal presence and selection of Cloud Services.  But, I'll tell you something that is not widely known, just because something is well marketing and looks like a good price doesn't mean it is built to last and to handle stress.  An example of this I've noticed is ski jackets that get used by ski mountain staff.  They wear the jackets multiple times a week, wash every week (we hope), putting more stress on a jacket in a month than most would put on their ski jackets over 5 years.  Quickly, they learn what will really last or not, and how it is priced and what brand is many times irrelevant.  They learn to look at the material, construction, and quality of the jacket and where their past jackets have broken.  fyi, AWS breaks in various areas that developers run into which can frustrate the hell out of them.  I know this first hand because my friends have gone through the hell of finding where AWS breaks, and had to make the choice to build the services themselves.

So, let me walk through Urs's keynote and point out some of the cool infrastructure things.  Note: I mention in the title than I'll compare to AWS.  Well I threw that up to get your attention, but actually there is very little out there if any that discusses AWS infrastructure.  It is like a retailer, it is all about price, selection, and selling to the target audience.

If you want to see the presentation go to about 21minute mark in this video.

Here is Urs's title slide.

NewImage

The next slide, Urs's discusses the physical world of data centers to support the cloud.  It's not a bunch of fluffy stuff that scales infinitely.  It is built on physics.

NewImage

 

Then slides showing the physical infrastructure that Connie Zhou documented in her pictures.  There weren't any new pictures that most of you haven't already seen.

NewImage

The environmental message is delivered.

NewImage

Google's announcements over the past year of expansion.

NewImage

Besides building data centers, Google runs their own network with their own sub marine cables

NewImage

The network spans the world and at some point will most likely reach Africa, Middle East, and India.

NewImage

 

One of the thing Google does is it thinks of its SW as infrastructure.  Urs reviews the history of the Google SW infrastructure.

NewImage

At about the 30:24 mark Urs discusses the obsolescence of features to learn from the past and make things better, focusing on quality and performance.  Sounds like my ski jacket story above. :-)

163 improvements are listed over 12 months.

NewImage

Customer wins are discussed on the platform.

NewImage

To disrupt the business model of an AWS, Google has added sub minute billing.

NewImage

In the spirit of a little green server, Google has a micro VM of only 0.6 GB.

NewImage

Getting Cloud to be useful many times requires integrating with on data center services in the company's data center if you can have an encrypted VPN connection.

NewImage

Persistent disks are useful, but the standard is 1TB.  Google has announced 10 TB.

NewImage

Platform infrastructure is great, but what good is it if you can't develop apps.

NewImage

To meet the needs of Information Security, Google Cloud Service are ISO 27001 are certified.

NewImage

If you want to see an app to build at the 45:00 mark you can see a demo of building an app.

In building applications this is what AWS has tons of content on. 

NewImage

 

 

 

 

What's the rest of story behind Adobe's Business Catalyst move to AWS

Werner Vogel proudly tweets.

honored to support RT : moving to I'II never build another datacenter again

Here is the Adobe post on its move to AWS.

In the coming months, we will be focusing our attention on transitioning the Business Catalyst platform from our existing data centers over to this new cloud infrastructure. Based on the Amazon Web Services (AWS) cloud computing platform, this infrastructure will ensure increased stability, reliability and scalability as the number of sites hosted with Business Catalyst grows into the future. If you are familiar with AWS yourself, you’ll also know this means even faster load times and higher quality of service for visitors to your client sites.

Turns out there is more behind this story.

I'll write a future post on this story as there are some good lessons behind the move.

The next Cloud Feature - Frequent Flyer program to lock you in

I am participating in a Route to the Cloud webinar tomorrow and one the points I want to make is whether you want to be cloud independent or not is an important decision.  What fool would want to be locked into a cloud?  In fact GigaOm's Barb Darrow just posted on the issue of lock-in preventing movement to the cloud.

Fear of lock-in dampens cloud adoption

SUMMARY:

Data portability — the ability to move your information between clouds (or in and out of clouds) with relative ease — is a key concern of companies considering a cloud move.

It’s become a truism to say that data is the new gold –but that doesn’t mean there are easy answers about where to store this gold. For now, many corporate customers will hold back on full cloud computing adoption until they’re convinced that they can move their data off a given cloud as easily as they put it there in the first place. Face it: fear of vendor lock-in is not limited to the on-premises IT world and it’s time enlightened vendors get this problem in hand.

What would motivate people to accept a lock-in?  Frequent Flyer program.  The human behavior to get points is ingrained in people.  

Can you imagine if AWS launched a point program for the amount of Cloud Services used and gave program owners the ability the redeem points for Amazon.com merchandise?  Users would then have the incentive to have larger AWS bills and loyalty to a cloud provider is common.

No one in the cloud service has a frequent flyer program, and this probably will never happen.  Note: this is a what if AWS launched a frequent flyer program, not they have.  

This may sound crazy, but we have all seen people who go through extreme lengths for Frequent Flyer Points.  You cloud kind of do this already if you set up a Amazon.com Rewards Visa card.  If you spent a $1,000/month on AWS that would be 3 points/$1 = 3,000 points / 100 points/$1 = $30.  You spend $10,000 a month on AWS.  Not hard to do. You get $300 a month in your account.

Companies tried to take away an employees frequent flyer points, but that didn't work.

Welcome Amazon.com Rewards Visa Cardmember


Earn points on every purchase*

  • 3 Points per $1 spent on Amazon.com
  • 2 Points per $1 spent at gas stations, restaurants, drugstores, and office supply stores
  • 1 Point per $1 spent on all other purchases

Redeem your points instantly at checkout with Shop with Points

You can use the points you earn to buy the stuff you want right at checkout.

  • See your points automatically
  • Pay for all or part of your purchase - there's no minimum
  • Every 100 points = $1.00 towards your purchase

Google influencing Amazon.com's server purchasing, and many others

It is accepted that the big Web2.0 companies don't buy what the enterprise guys buy.

Wired has an article on how Amazon changed its server purchasing to follow Google.

Pinkham was struck by how different the machines looked — and how hot they were. Even then, Google was running its website on dirt-cheap, stripped-down servers slotted into extremely tight spaces. They didn’t even have plastic cases.

“They were clearly not your average Dell, HP, IBM servers. They were white box machines, very densely packed. They weren’t in containers. They were just blades jammed into these custom racks,” remembers Pinkham, who went on to lead the team that built the Elastic Compute Cloud and now runs a cloud software startup called Nimbula. “And I remember a lot of heat coming off them — an indication of a lot of concentrated power.”

This isn't really new to most of you, but it is nice to have a Wired article tell a story to the rest.

Economist Article on the fight between Google, Apple, Facebook, and Amazon

The Economist has a great article on the competition between Google, Apple, Facebook, and Amazon.

The article closes with an excellent point that the regulators could change the game rules.

The Others

Watchdogs in Europe and America have been looking into accusations that Apple has colluded with some publishers to break Amazon’s grip on e-books. And they have been scrutinising Google too. Some companies, including ones with links to Microsoft, have accused the search firm of unfairly promoting its own services, such as Google+, in search results. They also claim that it uses content from competitors without permission, and that it has struck anti-competitive deals in search advertising. The firm is under fire for allegedly using smartphone patents to stifle competition. Google’s legions of lawyers have been battling these charges.

Their lordships Page, Cook, Zuckerberg and Bezos thus need to map a course for their respective firms through dangerous legal and regulatory territory. At the same time they have to avoid being distracted from fighting their rivals; the mad emperors of Microsoft lost a lot of ground by taking on the inhuman might of the Department of Justice. And the shareholders, hungry for returns in a moribund global economy, need to be kept happy.

A king who pulled all this off might claim the throne by right; but his chances of being more than first among equals, or of a lengthy reign, would be slim. As in Westeros, these battles and plots promise many more sequels and series.