/. recently saw an interesting post about The Decline and Fall of System Administration. One of the arguments for this (perceived? real?) phenomenon is that server virtualization makes it easier to re-image or restart a server rather than troubleshooting it when a problem arises.
At Eucalyptus Systems we create and think Cloud Computing, so it is natural for me to question if Cloud Computing is involved at all in the above discussion. As you know virtualization != cloud (see Learn and Rich's blog). On the other hand, Cloud Computing makes it even easier to restart a virtual server (one euca-reboot-instances away available to all Cloud Users). So are IaaS and Eucalyptus partly responsible for the fall of System Administration?
It's interesting to note that just before the slashdot post, we published our thoughts on IT roles in the era of the Cloud (see Cloud IT Roles). With our article fresh in my mind, I set myself to answer the above question. In the Cloud IT Roles article, there are quite a few roles that have the capacity or the ability to work with instances (or "virtualized server" in the parlance of the original post). For example the Cloud User can start and stop instances at will, as can the Cloud Application Architect and of course the Cloud Administrator.
The depth of knowledge required to perform these roles is not the same, but also different are the consequences of their decisions (i.e. reboot an instance vs troubleshooting). The impact of a Cloud Administrator's decision to reboot key components of the cloud infrastructure (and following the original post down the path of a lengthy reinstall), can be substantial since all Cloud Users and End Users of that installation might lose access to their resources for a significant amount of time and may possibly lose unsaved work. Contrast this with the decision of a Cloud User to reboot her own instance, the effect of which is most likely felt by just one person—the Cloud User herself. Then there is the middle of the road example, where a Cloud Application Architect responsible for an application decides to reboot the instance(s) hosting the application. In this case, all End Users dependent on that application will be impacted by the decision, but no other application or part of the IT structure will be touched.
It is my opinion that while more people will have access to 'admin' account and capabilities, the Cloud will naturally enforce (limit) the impact of their decision to the resources allocated to them. In this respect, while a Cloud User is technically a super user of the instances she starts, if she uses images created by a Cloud Application Architect, she doesn't need to have the skills of a System Administrator. In this scenario, it seems perfectly normal to reboot (restart) a server since the Cloud Application Architect will have designed the images to be rebootable without loosing data or having any adverse effect. In short, the roles implied in Cloud Computing delineate clearly their own skills and responsibilities.
So is this the Decline and Fall of System Administration or the Rise of Cloud Administration?