How Amazon Swooped in to Own Cloud Services
bloomberg.com
bloomberg.com
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.
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.
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.
As far as I can tell, Amazon were the first firm to actually execute and deliver.
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)
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.
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.
That person was probably Seats.
"As of April 2015, DigitalOcean was the third largest hosting company in the world backed by 154,000 Web-facing computers..."
http://techcrunch.com/2014/03/06/digitalocean-raises-37-mill...
Edit: I missed the most recent round of investment. Looks like they've raised roughly $90M in 2014. Impressive indeed
http://www.forbes.com/sites/benkepes/2014/12/09/digitalocean...
People are building companies on their infrastructure. One example is Cloudways.
The've raised ~$92M (https://www.crunchbase.com/organization/digitalocean) so 92,000,000/154,000 ~= $600/"machine" if it's a physical machine and they spent almost their fundraising on hardware.
Seems like spending almost all of your fundraising on hardware could happen for a cloud computing company.
So could be 154,000 physical machines?
Last summer I went to an AWS Summit in NYC and it was clear that AMZN was trying to reach management-level people but was mostly getting techies. The crowd was blatantly disappointed in Zocalo and the lines to get into the more technical talks were incredibly long so I went to more of the business track events.
I flagged this article because it is pure propaganda, self serving and right out of the mouths of Amazon's PR staff. This is how they operate. This is how a retailer whose management has no understanding of technology and who abuses their engineers is made to be perceived as a "tech company". They have been engaging in wall to wall propaganda for the past 10 years, using techniques pioneered (and possible the same PR firms) by Microsoft when they wanted to change Bill Gates from the perception that he was evil incarnate into the perception that he was a good guy. (Which worked amazingly well. People seem to have forgotten how he set back technology 10-20 years. Hell, 20 year olds today were just being born during the height of his reign of terror.)
If this story about Sun is true, it shows just how much damage the US government does to the economy with "regulation". But since there rest of the article is nonsense who knows.
The claim that Amazon has a 7 year headstart is nonsense. The claim that Amazon "turned cloud computing into a mainstream phenomenon" is nonsense.
By the way, Amazon's only service in 2006 was S3. They claimed (it was a lie, and I know this because I worked at Amazon at the time and was amazed at the boldness of the lie when I saw the press release) that S3 used "the same machines that run Amazon.com!" Amazon didn't seek out to build a business with cloud hosting-- this is just another initiative thrown against the wall to see if it would stick. (Remember A9? Remember the literally 100+ businesses they have launched and shuttered over the past decade? The Fire phone is an example of this. They waste a lot of money just building crap to improve their image, make them seem like a "Tech company") S3 was actually innovative-- shows what can happen when management at Amazon takes their eyes off of the ball and lets engineers have a little space. (unusual at that company.)
What amazon is good at is marketing and particularly to the Hacker News crowd, I'm sorry to say. Objectively if you look at AWS it's a bad deal all around. The technology isn't that great, it's terrible when it comes to reliability, and the pricing is not very good either. Further, they lock you into their APIs which means you have to spend a lot of effort to leave. Yet whenever these things are pointed out here on HN, people say "yeah, but who has the time or money to run their own servers!!11!".
As if they was the only alternative, as if there weren't lots of major competitors from Google and Microsoft to Digital Ocean, Rackspace -- which will give you even more high touch services-- and a million mom & pops. Every feature in AWS was offered by somebody, at scale, before Amazon, and usually without a proprietary wrapper around it. In fact, most of this stuff existed in one form or another in the early 1990s.
Amazon only "Dominates" in the minds of the PR flacks who write these articles and the proof is in the first sentence of the next paragraph: "On April 23, Amazon plans to disclose the numbers for its cloud business for the first time"
This whole thing is a shell game akin to the one Microsoft played in the 1990s where they bought up all the media, and every writer out there whose audience included CIOs and paid them to say that Microsoft had won, that Apple had no market share, and by doing so persistently for several years (much like google is doing now with android) they convinced people that Apple had less than %5 marketshare. This at a time when Apple had %25 market share! This killed many software projects-- I remember being told "there's no market for our application on the mac" during that era by marketing guys who knew nothing of the real stats, only what they read.
Meanwhile people think android is winning based on ZERO sales data, merely a constantly and loudly repeated propaganda making the claim. Apple publishes audited actual sales numbers every quarter. No maker of android phones (that I've found) does so.
If you think android is dominating in sales you're believing propaganda and have no proof, and probably no evidence. (and don't get me started on "activations". At one point Microsoft was "selling" and counting, windows for every mac installed in every school!)
I see the down votes are coming in, I know to expect this because I'm daring to criticize a "major tech company" that's often mentioned along with Facebook, Google, Microsoft and Apple.
But my point is, we should not be so susceptible to propaganda.
This article is just pure propaganda.
What I'm taking from your comment, is that your down vote comes from being offended that I said something negative about something negative being done by a company you care about.
I upvoted your comment, by the way, even though I think it's wrong, because it was constructive, and illuminated your thinking to me, and I appreciate that. My point was pretty well argued in my comment, and thus, even if you disagree with it, I don't deserve a down vote.
The "down votes because I disagree" is literally saying "I want to silence those I disagree with, rather than go thru the effort of rebutting them". Literally on this site as you go negative, you go to lighter shades of grey, and eventually disappear.
Voting was great for dealing with spam, but alas, it's become enforcement of the ideological filter bubble.
Check out CloudKit, or Digital Ocean. AWS is really showing it's age. It's now just a Chinese List of confusing services, with just horrible documentation. (IMHO).
It's Monday at AWS, lets see: OK, today lets add: Cloud Replications Across Distributed Mobile Latency Image Systems Plugin Transient Architectures!
Great, throw up a link!
I'm sure that's what they are doing. Their services seem very similar to the Bullshit generator. Kind of spooky :-)
Damn, I just lost my: Amazon Elastic Transcoder! Holy shit, now what? OMG :-)
Just killed my AWS account, and this is what they sent:
This e-mail confirms that you have cancelled your access to AWS Unified Registration. Your access to following services is canceled: Amazon Simple Storage Service Amazon Elastic Compute Cloud Amazon Simple Queue Service Amazon SimpleDB Amazon CloudFront Amazon RDS Service Amazon Elastic MapReduce AWS Import/Export Amazon Virtual Private Cloud Amazon CloudWatch Amazon Simple Notification Service Amazon Simple Email Service AWS CloudFormation Amazon Route 53 AWS Elastic Beanstalk AWS Direct Connect AWS Storage Gateway Amazon ElastiCache Amazon CloudSearch Amazon Simple Workflow Service Amazon DynamoDB Amazon Glacier Amazon Elastic Transcoder AWS Data Pipeline Amazon Redshift AWS OpsWorks AWS CloudHSM URPProduct Amazon WorkSpaces Amazon AppStream Amazon Kinesis Amazon Zocalo AmazonCloudWatch Amazon Cognito Amazon Mobile Analytics AWS Directory Service AWS Key Management Service AWS Config AWS Lambda Amazon EC2 Container Service AWS CodeDeploy Amazon Machine Learning Amazon WorkDocs
Want static file store on Digital Ocean? You'll probably wind up using S3. Want a load balancer? Have to build one yourself with HAProxy (and deal with failover somehow). Want a managed database, queueing service, etc.? Nope.