The Charm of Centralization

Centralization

“All under heaven, what has been united long time, that must be separated, what has been separated long time, that must be united…” (分久必合 – 合久必分) This citation from the Romance of the Three Kingdoms recently came into my mind, a famous novel about China’s turbulent ancient history. Obviously: Centralization is an old concept (just like separation is).

Chinese character for "center, middle, China"

Some people argue that Cloud Computing is a step back to where we came from. Everything started with huge mainframes and dumb clients, then we moved towards the more egalitarian PC and client/server paradigm and now it seems to go back to the mainframe again, called “Cloud”, plus dumb clients like mobile phones and netbooks.

I believe this observation is correct. In fact, Cloud Computing differs from other concepts of distributed computing (like Grid Computing) in that it promotes single administrative domains. Everything is under the control of Amazon/Google/Microsoft/… Well, the provider-side benefits are obvious: lock-in. What are the consumer-side benefits?

Evernote

Evernote is a great tool (no affiliation whatsoever). They promote it as “external brain” because you can capture and store all kind of information in their Cloud – and find it again. The second part is pretty important. You can capture and annotate all kinds of things, such as Web sites, pictures taken with your mobile phone camera or audio clips taken with your smartphone mike. Pictures that are clipped with Evernote are processed afterward and text within the picture is indexed and thus made searchable. Neat!

What is the idea behind Evernote besides these cool features? Everything in one place, everything annotated and indexed. Evernote promises to sync with all kinds of devices, making it the core platform for collecting and organizing your information. The concept: Centralization.

SaaS

Back in 2006 Frederick Chong and Gianpaolo Carraro have written a great article about the advantages of Software-as-a-Service. The implications for business models boil down to 4 points:

• Pay for using software, instead of “owning” software
• IT complexity shift from consumer to provider
• Leveraging economies of scale (that are passed on to the consumer)
• Targeting the “long tail” of small businesses

Exchange the word “software” to “hardware” and you have a valid statement for Cloud Computing. The concept behind SaaS? Just like Cloud Computing: Centralization. Complexity is shifted to a centrally managed platform that can be administered and operated with lower costs per output unit and that leverages long-tailish business development.

Decentralized Centralization

Centralization has a bad reputation, it sounds like Single Point of Failure. Centralization is efficient but it comes with a few disadvantages that might lead to severe problems. In analogy to the Internet (network of networks) the future of Cloud Computing might be a Cloud of Clouds. Within a single Cloud there is a single administrative domain (efficient). Users can tap into different Clouds, just like Multihoming, and shift data and applications from one Cloud into another one (robust). CohesiveFT shows a first step towards this direction: VPN-cubed Clouds.