We had an excellent brainstorming meeting on Personal Clouds at IIW yesterday. About 20 people showed up to discuss their views on Personal Clouds: what they are and aren’t, what they need to do, will do in the future, and what they can’t and shouldn’t do. I was really surprised by how much agreement there was, and how much we were on the same page even about some rather advanced details.

Personal Clouds are coming, watch my words.

Here are my notes from the meeting:

Personal Clouds are to cloud computing what the Personal Computer was to (mainframe-based) computing at the time: in the mainframe world, a central group of people (the computer operators) would deliver the apps they choose, with the data they controlled, with their terms of service, to users who had to use whatever was given to them. The Personal Computer with personal productivity apps such as spreadsheets was the counter-trend, which put all aspects of computing in the hands of the users, who could add and remove hardware, create, delete and modify data at will, run whatever apps they chose on whatever data they had.

Cloud computing today is largely the same as mainframe computing was back then: a central group of people (the operators of SaaS apps such as Facebook or Salesforce) decide which features to give to users. They can change features at will without requiring the consent of their users (e.g. Facebook timeline, no way of not using it), who control the user’s data and often make it hard to move the data somewhere else, or even access it from somewhere else (e.g. no Google crawling of Facebook) and subject to their terms of use. E.g. Facebook or Google can unilaterally ban users and delete their data, which they have done.

Personal cloud computing puts control back in the hands of the users, but this time in a cloud / networked environment.

There are three major parts of personal clouds:

• data
• capabilities (aka apps)
• terms and governance

For the cloud to be personal, the owner of the personal cloud must be able:

• to choose and remove the apps they run on their personal cloud
• to control who does and does not get access to the data on the personal cloud.
• to process data created with one app with another in a similar way as files on a PC may be opened by apps from a different vendor (something not possible with SaaS today)
• to set the terms of use themselves of the personal cloud
• to move the personal cloud from one host / infrastructure / hosting provider to another if needed (e.g. from an Amazon cloud server to a Rackspace one)

Cloud computing features such as replication, and high-availability features should be available. Backup must be available.

Contrast to what some people called “personal data centers” — set of computing resources some people (usually techies) have control over at home or work. This wouldn’t be cloud computing because in cloud computing, somebody else takes care of failing hard disks etc.

Personal clouds interact with each other as peers. So we don’t believe that Dropbox etc. are personal clouds. They are just a service that might be used by somebody as one component of a personal cloud.

Personal clouds require persistent identifiers / identity that can be allocated and asserted in a decentralized manner.

May be difficult to communicate because many people do not understand the difference between data and apps.

It’s a matter of control and portability.

Today mobile devices are tethered to service providers. This creates a dependency on service providers that is undesirable for personal clouds. Of course it is a reality that some technologies today are only available tethered, e.g. iPhones.

There was 90%+ agreement in the room that 5 or 10 years from now, most people will have personal clouds.