Internet and Businesses Online

Cloud-computing is the buzz word of 2008. One of the first commercialized cloud services was provided by Amazon, offering EC3 (Elastic Computation Cloud) and S3 (Storage Cloud). To consume a service like S3, Amazon provides a simple web services interface through which a user saves and retrieves data. Where the data actually gets stored is only transparent to Amazon. For the user it's somewhere in the cloud. A customer only pays for the bytes stored and transferred from and to the cloud. It's a true on-demand system which is capable to scale to any need. The ease of use and low cost of entry attracted many companies, especially web 2.0 startups.

The growing demand is also having its downsides. An outage of Amazon or Google takes down thousands of companies at the same time. A company can't do much but wait until the cloud comes back up again. This scenario is comparable to a power outage. An emergency generator could things keep temporarily running. In case of a cloud outage data is completely inaccessible.

Don't put all your eggs in one basket. Strategically it makes sense to use cloud-computing but from different independent providers. For document management there are different providers like Expresso, Zoho, etc. CRM systems are provided by 37Signals, SalesForce.com, SugarCRM. Choosing different providers for different work tasks lowers the dependency and risk of outages. Many companies provide access to its service from inside other clouds like SalesForce.com, Facebook, OpenSocial. E.g. Does SalesForce go down, a user could still access documents from Expresso outside SalesForce.

IT has always transitioned between centralization and decentralization. Cloud-computing is nothing else but the next evolutionary step towards centralization. History has shown that too much centralization makes a system vulnerable and hard to manage. In 10 years from now, will everybody talk about raindrop computing?