Data centers load is basically server work load and cooling load associated with it. In literature papers, there are ways to model the IT load (delay tolerant and delay sensitive) and also cooling load, but how can we use that model to find uncertainty of the load. Is the uncertainty assumed by specifying the range of demand, or can be solved as a stochastic optimization considering different scenarios?
If you're looking for a market-based mechanism, you may look here:
Bapna, R., Das, S., Garfinkel, R. and Stallaert, J., "A Market Design for Grid Computing," INFORMS Journal on Computing, Winter 2008, 20(1), pp. 100-111.
* During my literature review, it appeared to me that two rules are applied by service providers :
1- Provision a certain amount of load (usually around 20 - 25%). Therefore, a part of uncertainty can be absorbed by that (but not totally).
1- If even with the provisioning a machine is overloaded, then kill the service (VM) with less priority.
* Regarding the way of modelling it. I would say that datasets are already including this uncertainty (load changing over the time). Have a look at the google cluster trace (it contains infos about VM loads variations over their execution time): https://code.google.com/p/googleclusterdata/
* If you would like to introduce uncertainty, this could help avoiding over provisioning (Good). However, it will include a large failure risk (bad, but reducing it can be your objective).
* Another issue would be the scale of your problem as it is usually large scale.
you may have a look at this paper :
https://www.researchgate.net/publication/258835771_Decision_making_under_uncertainty_in_energy_systems_State_of_the_art
Article Decision making under uncertainty in energy systems: State of the art
- Badges
- Science topic
- Similar topics
- Engineering
- Computer Engineering