5 ms·

I didn't remember the Sun initiative specifically, but it sure sounds like they never got close to bringing anything to market. So I'm not sure if it's fair to

I didn't remember the Sun initiative specifically, but it sure sounds like they never got close to bringing anything to market. So I'm not sure if it's fair to characterize Amazon's position of being the first to market by many years as "swooping in". Amazon has essentially defined the category. But I think the key to their success in something that ultimately ended up being completely revolutionary was in starting small and unpretentiously. They didn't promise that SQS or S3 or EC2 would change the world. They didn't offer every feature imaginable up front. They grew their services piece by piece, and layer by layer, and they continue to do so every year. The biggest surprise to me was always that it took Google so long to get into the same game.

mitchell_h 12y ago
Sun's initiative was on the market. You could(I did) buy time on their grid. However, to characterize it as something that would have or competed with amazon is just wrong. Sun's offering was much more of a pre-google compute engine. There were other difference in sun's offering was push more towards bulk compute and less end user applications. I'm pretty sure Sun also tried to add things that would entice some end user applications after launch.
captn3m0 12y ago
Google decided to instead focus on App Engine, and even that was launched in 2008 (and out of preview in '11).
ghshephard 12y ago
Amazon was definitely not first to market - there were a large number of cloud providers (Loudcloud being just one of them) that tried (and failed) to do what Amazon eventually succeeded in doing. The idea was pretty similar though - just build simple services that Customer Could build applications on top of. Loudcloud had its first customers in 1999, 7 years prior to Amazon. Totally agree with you - what Amazon did was start off small, and unpretentiously, and just keep growing/iterating from there. LDCL went the total opposite approach, spent 10s of millions of dollars (I remember a time in 2001 when we were buying Sun E4500s like popcorn @$100K+ each) in the first couple years, hired 500 employees, opened up data centers all over the world - and kind of fell down hard when the dot com market imploded.
- wtbob 12y ago
  I remember Lou Gerstner predicting the rise of utility computing way back in 2002[1]. It was a great idea, but IBM were simply unable to execute on it. As far as I can tell, Amazon were the first firm to actually execute and deliver. [1] http://newsoffice.mit.edu/2002/gerstner http://newsoffice.mit.edu/2002/gerstner
  - ghshephard 12y ago
    It's tricky - "Utility Computing" can mean many things to many people. I think the subtle challenge is in understanding the difference between "Cloud Services", "Hosted Services", "Web Hosting", "Professional Services", etc... There were a lot of "Managed Web Hosting" companies back in 2000/2001, doing quite well. Logictier, Sitesmith, Mimecom, Loudcloud were some of the big ones. What Loudcloud tried to do was provide a platform that other companies could drop their application on top of. Loudcloud fought and struggled for the longest time to keep the users from having root on their servers, and installing whatever they wanted - until they finally gave up that battle, and just let the customers have at it. What LDCL never understood, (or, perhaps, it was just to early to deliver) is that the real, amazing scale came from having simple services that are 100% automated to procure, and then letting the customers procure and deploy those services, and then build their applications against those services. That is, provide scalable services, and let the customers worry about how to scale their applications to use them. A lot of the early LDCL DNA came from AOL, where it was observed that every time a business was placed on AOL, they would collapse under the flood of traffic that came at them. As such, it was believed that most businesses wouldn't know how to scale their system - but they could write their functional code. So the division was, "You write the application, we'll scale the rest" - If I understand what they do, it feels like Heroku ended up executing on LDCL's original vision. LDCL, in some ways, executed well, but on the wrong vision (or, once again, perhaps the timing was wrong). They wrote massive systems management/package management software, that would allow people to deploy code, software, on a large number of servers but never really got into the business of supplying "cloud services" the way we think of them now. Still, $1.6 Billon exit in 8 years on around $350mm raised - not too shabby. And talk about your pivots...In the middle of an economic (web) collapse no less. Ben Horowitz is quite the street fighter... (More details here: http://www.bhorowitz.com/the_case_for_the_fat_startup http://www.bhorowitz.com/the_case_for_the_fat_startup)
dba7dba 12y ago
Amazon had the luxury of borrowing engineering and $ from other side of their business to build that 'small and unpretentious' cloud. No other companies have the $ or engineering to pull it off. Nor the time to wait for it to grow. This reminds me of South Korea's chaebol and Japan's keiretsu business groups.
- billyhoffman 12y ago
  A good point, and probably helped Amazon's success. But this is no different than Sun. Have you seen the margin on SPARQ processors and machines? Sun had plenty of cash to bootstrap their grid computing effort. The problem for Sun was a Strategy Tax. Sun was in the business of selling big iron systems. Their grid effort was an inverse of this model: 1- Sun had to pay for/run the systems (or sub contract it) 2- Why buy a big system from Sun when you can just use what you need? 3- Sun's sales organization was not designed to sell utility computing to medium sized businesses.

mitchell_h 12y ago

Sun's initiative was on the market. You could(I did) buy time on their grid. However, to characterize it as something that would have or competed with amazon is just wrong. Sun's offering was much more of a pre-google compute engine. There were other difference in sun's offering was push more towards bulk compute and less end user applications. I'm pretty sure Sun also tried to add things that would entice some end user applications after launch.

captn3m0 12y ago

Google decided to instead focus on App Engine, and even that was launched in 2008 (and out of preview in '11).

ghshephard 12y ago

Amazon was definitely not first to market - there were a large number of cloud providers (Loudcloud being just one of them) that tried (and failed) to do what Amazon eventually succeeded in doing. The idea was pretty similar though - just build simple services that Customer Could build applications on top of. Loudcloud had its first customers in 1999, 7 years prior to Amazon. Totally agree with you - what Amazon did was start off small, and unpretentiously, and just keep growing/iterating from there. LDCL went the total opposite approach, spent 10s of millions of dollars (I remember a time in 2001 when we were buying Sun E4500s like popcorn @$100K+ each) in the first couple years, hired 500 employees, opened up data centers all over the world - and kind of fell down hard when the dot com market imploded.

wtbob 12y ago
I remember Lou Gerstner predicting the rise of utility computing way back in 2002[1]. It was a great idea, but IBM were simply unable to execute on it. As far as I can tell, Amazon were the first firm to actually execute and deliver. [1] http://newsoffice.mit.edu/2002/gerstner http://newsoffice.mit.edu/2002/gerstner
- ghshephard 12y ago
  It's tricky - "Utility Computing" can mean many things to many people. I think the subtle challenge is in understanding the difference between "Cloud Services", "Hosted Services", "Web Hosting", "Professional Services", etc... There were a lot of "Managed Web Hosting" companies back in 2000/2001, doing quite well. Logictier, Sitesmith, Mimecom, Loudcloud were some of the big ones. What Loudcloud tried to do was provide a platform that other companies could drop their application on top of. Loudcloud fought and struggled for the longest time to keep the users from having root on their servers, and installing whatever they wanted - until they finally gave up that battle, and just let the customers have at it. What LDCL never understood, (or, perhaps, it was just to early to deliver) is that the real, amazing scale came from having simple services that are 100% automated to procure, and then letting the customers procure and deploy those services, and then build their applications against those services. That is, provide scalable services, and let the customers worry about how to scale their applications to use them. A lot of the early LDCL DNA came from AOL, where it was observed that every time a business was placed on AOL, they would collapse under the flood of traffic that came at them. As such, it was believed that most businesses wouldn't know how to scale their system - but they could write their functional code. So the division was, "You write the application, we'll scale the rest" - If I understand what they do, it feels like Heroku ended up executing on LDCL's original vision. LDCL, in some ways, executed well, but on the wrong vision (or, once again, perhaps the timing was wrong). They wrote massive systems management/package management software, that would allow people to deploy code, software, on a large number of servers but never really got into the business of supplying "cloud services" the way we think of them now. Still, $1.6 Billon exit in 8 years on around $350mm raised - not too shabby. And talk about your pivots...In the middle of an economic (web) collapse no less. Ben Horowitz is quite the street fighter... (More details here: http://www.bhorowitz.com/the_case_for_the_fat_startup http://www.bhorowitz.com/the_case_for_the_fat_startup)

dba7dba 12y ago

Amazon had the luxury of borrowing engineering and $ from other side of their business to build that 'small and unpretentious' cloud. No other companies have the $ or engineering to pull it off. Nor the time to wait for it to grow. This reminds me of South Korea's chaebol and Japan's keiretsu business groups.

billyhoffman 12y ago
A good point, and probably helped Amazon's success. But this is no different than Sun. Have you seen the margin on SPARQ processors and machines? Sun had plenty of cash to bootstrap their grid computing effort. The problem for Sun was a Strategy Tax. Sun was in the business of selling big iron systems. Their grid effort was an inverse of this model: 1- Sun had to pay for/run the systems (or sub contract it) 2- Why buy a big system from Sun when you can just use what you need? 3- Sun's sales organization was not designed to sell utility computing to medium sized businesses.