Last year people from a big IT Blog asked me to write an article about cloud computing for their site … but it actually never got published. Here it is:
I always wanted to own the things, that I use. Movies, Music, Gadgets and lots of computer power. Perhaps it is a menfolk thing to hunt the booties and take them to your cave. Perhaps it is just stupid me, but believe me I am a collector.
Knowing this you would be very surprised if you enter the doors of my office.
Today the total amount of all capital assets of my company rest in some MacBook Pro Laptops and a Coffee Machine. Next week a camera crew is doing a portrait on our company and they asked me what they would be able to film in the office. And I had to say: Nothing, but we can take a walk at the beautiful waterside near my office. This is one major effect of a technology that is called cloud computing. You own nothing and yet everything keeps on working – and you are able to take a walk at the waterside (if available).
“The cloud” is really rather a dumb name for a revolutionary technology isn’t it? Because a cloud is flawy, nebulous and intangible which in other words mean, that you will never be able to describe it in a reasonable way.
Some abbreviations try at first glance to irradiate the dark, but at the second look SaaS, PaaS and IaaS are much more the sound I make, when I have heavy snuffles than any form of explanation onto what the cloud is.
In early web development stages IT people tended to draw a cloud in their diagrams when they meant “the internet”. There were arrows going into the cloud and coming out of the cloud and that meant “we don’t exactly know whats going on here”. So cloud computing is in that terms computer science where the IT technicians themselves don’t really know whats going on. And from my perspective today this is completely true. But it is rather a relief than a thread!
The cloud turned itself into something powerful much more like the black cloud-monster in LOST (without all the murder) than the fluffy “this-looks-like-an-elephant” cloud.
Our company (schnee von morgen webTV GmbH) builds up TV stations for customers like DER SPIEGEL, Discovery Channel and dctp and in the last years where everything turned “webby” we started to design and instantiate webTV projects.
The combination of films and the web is foremost very expensive with a strong leaning towards being commercially impossible. There are so many different codecs and platforms, that transcoding itself is one major doorkeeper. When the films are transcoded, worldwide delivery for a major broadcaster is an incredible complex problem – and it is very expensive.
Doing all this on a self owned IT infrastructure is a great idea if you just won the lottery twice in a row, but if you haven’t this is not a reasonable approach.
The TV business has such a high volatility in the production and the consumption of content that you have to have a really big IT infrastructure to cover the peaks. Which in effect means that you pay for boredom most of the time. Or in other words: Nothing is so expensive as the idle time of your infrastructure.
To solve this problem a high amount of virtual imagination is necessary. Because the first thing that solutions in the cloud do is to take things away that have become dear to us in the last years like:
– The feeling of standing in the server room at night with millions of LEDs blinking and the loud noise of the air conditioning.
– The ability to scream at somebody irrespective of the error that occurred.
– Solving problems “the old way” which in most cases mean: by fax machine.
But if you are willing to disclaim a little and try new ways of problem understanding and solving then cloud computing is going to enable things you might today think of being impossible.
Cloud computing is about sharing resources and that the whole is greater than the sum of the individual parts. Basically it is a beautiful idea.
But what about reliability?
“The major difference between a thing that might go wrong and a thing that cannot possibly go wrong is that when a thing that cannot possibly go wrong goes wrong it usually turns out to be impossible to get at or repair.” Douglas Adams, Mostly Harmless English humorist & science fiction novelist (1952 – 2001)
This is so true. And it is a allegory for the development in computer technology as I see it. For most part of the last 10 years I tried to construct unbreakable systems. This turned out to be very complex, very expensive and very very impossible.
Today we at schnee von morgen are proud to build systems that break all the time. The trick is not perfection, but rather a frowzy way of best practices and imperfection.
Its about teaching computers to behave like we want other humans to behave if things become difficult. To keep an eye on each other, to stand close in times of trouble and to call for help if needed.
Under these preconditions it is possible to construct self-healing and extreme stress resilient IT infrastructures.
Tradeoff is you are going to loose a certain amount of control. The high amount of virtualization that is required to share IT resources “in the cloud” leads to obscure intransparency if you want to understand your IT infrastructure as a whole.
There are times where I am not able to say how many servers on how many continents are currently running for my company and what problems they are currently solving. That is mainly because they organise themselves and the servers are only reporting solutions or problems.
It’s really a new way of thinking because you have to accept certain conditions as given. It’s like math were you have to accept that 1+1 is 2 and if you begin to think through this small precondition a whole science comes out.
In terms of cloud computing its about accepting a certain amount of loss of control which must lead to reliance and trust into one or more providers.
You will still be able to control results, but you will loose a certain amount of control of the underlying processes.
And you will loose that poor guy you shouted at every time when something IT-ish went wrong. Actually there are some very expensive placebo support contracts available, but they do just that: they give you a very expensive person to shout at 24/7. That doesn’t mean, that in case of a infrastructure failure in the cloud this person is able to explain to you what went wrong or to prioritize your needs. That is because thousands of other companies that work “in the cloud” will have the very same problems. And either the problem is solved for everyone, or for no one
So you could just save the money for that poor guy and take a walk. Being assured that everything possible is done to fix the problem. And next day in office you start designing a workaround for yesterdays error and teach your cloud to heal itself in case the error happens again.
I am actually happy about every real life error that occurs, because every time something unpredicted happens I learn a lot. It’s about thinking of things you haven’t thought of you haven’t thought of. And every real live trouble makes the infrastructure stronger and more reliable.
And if you think at this moment: I will not be able to trust in something inscrutable like this.
1.) How does your car work.
2.) How does cell division work.
3.) How does your tax computation work.
Because your life is going to depend on 1 and 2 – and your business ideas are definitely affected by 3. All in all trust is not such an uncommon thing for us. It is only uncommon in terms with computers.
If this article makes you think “why don’t I too take part in this cloud-thing” … presumably you already do. Facebook, Google, Flicker, YouTube, del.icio.us, Twitter, webmail – ever used or heard of any of these? Big parts of our digital life are already sourced out to the cloud and this process is having a huge momentum.
It seems to be the natural movement in IT business to share resources to reduce costs. I’m not saying, that cloud computing is risk-free or that any kind of self immanent benefit comes with the cloud. There are new risks that come with shared cloud infrastructure independent from the type of cloud usage: shared infrastructure, shared platforms or shared web services come with new risks. We haven’t even got a full overview of the types of new risks that shared IT resources do. But I like to say “We cross that bridge, when we reach it.”
To close the circle: I’m going to take a walk at the waterside now. That is because I know, that there are currently no problems in our cloud and furthermore I have trust that upcoming problems will be solved. The beauty about this is, that I can concentrate on real problems like: How will we watch TV in the future? And I don’t have to worry about infrastructure anymore.