The Dawn of the Personal Cloud
Simple founder and former Twitter engineer Alex Payne has released sovereign - a set of open source Ansible playbooks to create a personal cloud. This allows users to move away from large cloud companies and onto services that are under their own control.
The point of a personal cloud is to put data under a user’s control rather than in the hands of a hyperscale service provider. Payne cites concerns over privacy, open standards, product deprecation and unwanted integration with social network features as reasons to move away from providers such as Google.
Sovereign brings together services for email, calendar, web hosting, Internet relay chat (IRC) and virtual private network (VPN) using a range of popular open source applications. It also includes components for security (file encryption, firewall, intrusion prevention), monitoring, and backup. All of these pieces could have been installed by hand, but the use of a DevOps tool makes everything more convenient. In this case Ansible was chosen over tools like Chef or Puppet as it works with remote machines over SSH rather than needing a local client to be installed. The outcome might not yet be quite as convenient as signing up to a cloud service provider, but it’s a lot more convenient than installing and integrating multiple applications and services by hand.
Personal application clouds like sovereign can be run on Infrastructure as a Service (IaaS), virtual private servers (VPS), private servers or virtual machines. Using IaaS or a VPS involves some degree of service provider trust when it comes to security and privacy, whilst private hosting puts security and privacy firmly in the hands of the user. Internet connectivity and cost of storage can also be differentiators.
The need for some kind of server puts personal clouds out of the reach of everyday users, meaning that for the time being adoption will come from tech enthusiasts and small companies. If our technology follows the usual Moore’s law trend then over time that server component should scale up and out, becoming something that can serve the needs of individuals (as a set of distributed and virtualized components on phones etc.) and the enterprise (on a scalable private cloud). For the enterprise this will mean a shift away from specialists who maintain applications and infrastructure and a move towards pre-integrated and auto-scaling solutions.
Productivity applications are just the start for personal clouds. As users adopt devices that generate video streams (such as Google Glass), and more devices connect to the Internet of Things (IoT) there are cases to be made for user control of data and a shorter network path to storage. Startups like CTRLio are emerging to exploit the opportunities in this space, and the NoBackend idea is gaining traction - where users choose which service will store their data. (QCon San Francisco 2013 has a track covering NoBackend). Perhaps the cloud skeptics like Cory Doctorow and Richard Stallman were right with their cost and control objections, or perhaps we’re seeing yet another industry shift from a centralised model to a distributed model.
"...Using IaaS or a VPS involves some degree of service provider trust when it comes to security and privacy, whilst private hosting ...."
IaaS and private hosting are not mutually exclusive. It seems the author is confused between the types of clouds and deployment of clouds.
Types: IaaS, PaaS and SaaS
Deployment: Private, Public, Hybrid and Community.
What the author is describing seems to fall under SaaS - as long as it adheres to the principles of the cloud (scaling, provisioning ...etc). The services/products described in and of themselves do not mean "cloud".