Bio not provided
I agree on all of your points as to tiering but must add that for the system to be truly effective, you must unleash the policy engine from the derived model and let it substantially adjust the system based on the best interests of the system...all within the constraints you impose by the local and global rules you write. An example is that after you do all of the analysis of a composite application (multi component; DB server, app server, web server) and it should be holistic and end to end analysis (its Quality of Experience or QoE) and initially assign it a tier or priority, you then let the system then use the applications behavior as a modifying force from there on. You might write a local rule for that application that sets a minimum tier but otherwise, the system sets the tier and takes action based in that and the rest of the behavior of the system.While, again, I agree that this can be termed elasticity, it goes far beyond just taking into account load on servers but addresses the non-technology aspects of applications such as political investment, compliance regime, time (clock and calendar) as well as qualitative characteristics assigned by the system owners.
3 years ago on The Hypothermic Cloud Infrastructure: Maintaining the Blood Flow to Tier 1 & 2 Apps