Jacques Mattheij

Technology, Coding and Business

Small really is good, the case for micro clouds

With some reluctance I welcome the coming of the cloud. It’s the logical next step, to offer services to a large number of users should be cheaper than to set them up for yourself due to economies of scale. We should all be better off.

But for now the prices aren’t that competitive, only in very specific use cases (I need 10,000 servers tomorrow and then I won’t be needing them for the next year or so) is there a very clear economic advantage.

And then there’s a distinct shadow side to this cloud stuff, which is that it goes against the whole philosophy that the internet was based on.

We were supposed to be ‘peers’, John Walker (formerly of Autodesk, hacker supreme and internet Wise Old Man ™) writes in ‘the digital imprimatur‘: “the Internet, properly used, could actually roll back government and corporate encroachment on individual freedom by allowing information to flow past the barriers erected by totalitarian or authoritarian governments and around the gatekeepers of the mainstream media.”.

Now, John Walker was mostly concerned with the firewalling of consumers, making it impossible for programs like his ‘SpeakFreely’ to communicate peer-to-peer. And then Skype happened, showing that it was a technical hurdle that could be taken.

But the main thrust of John Walkers argument stands, and it can be applied to consumers and to small corporations (service providers) alike: if we continue on the path to centralization that we are on today then we will erase a large portion of the potential of the internet.

Smaller is not just better, it is different!

Utility computing, while nice in theory effectively yanks us back to the days of the mainframe, only by a different name. Larry Ellison agrees, even if I think he has a bit of a different look on things because of the long term effect of the cloud on his bottom line.

Sure, there are huge differences between cloud computing and mainframes, mainframes almost without exception are fairly specialized beasts running specialized software, whereas the computers powering the cloud architectures are as simple and universal as they can be made, using commodity hardware and tons of open source software to drive them.

But that’s where the differences end. The similarities are conceptually speaking much larger, for starters you no longer have access to the hardware, you are beholden to some giant corporation (ok, maybe now it’s Google or Amazon, not IBM, Sperry-Univac or Burroughs). Each cloud implementation requires you to specialize to the interface of that particular service, effectively causing ‘lock-in’. Moving is expensive, and portability between cloud architectures is minimal at the moment. Standardization of the interfaces for the various cloud architectures would go a long way towards mitigating this.

Just like in the days of the mainframe, the cloud having issues is a major problem. The internet was meant as a secure network able to route around problems in its networking infrastructure. Routing all those lines to the same single-point-of-failure where the cloud happens to be hosted negates a lot of that robustness.

Just like with airplane crashes vs automobile crashes, airlines may be safer per passenger mile flown vs cars per passenger mile driven, but when they go down it is instantly headline news world wide.

To cut a long story short(er), I think there are some real drawbacks to all this cloud stuff, probably many of those still invisible, but it’s my gut feeling that we’re not on the right path with this.

Personally I’d like a completely different thing to happen. The hosting providers of today, whose business is ultimately seriously under fire from the big providers of cloud services should band together and launch their own version of the cloud. Let’s call them ‘micro clouds’.

A common interface that will work with every provider, that offers basic storage, database, CPU and bandwidth services. The customers are free to host where-ever they want because of the portability, ensuring real competition.

And Amazon, Google etc will not be able to intensify their grip on the internet any further than they already have.

I think if that were to happen it would be a good thing.

HN Submission/Discussion
If you read this far you should probably follow me on twitter: