Dear Researchers;
In energy aware VM scheduling, we tend to determine current server temperature or load before we activate a VM on that server. Is there a way we can predetermine the temperature change when the incoming VM is activated?
Thank you.
Hi there. I think those references would help you:
BOOK and Article
https://books.google.pl/books?id=9NFYCwAAQBAJ&pg=PA212&lpg=PA212&dq=measure+virtual+machine+temperature+load&source=bl&ots=fz4NS_QrSa&sig=lTr80WCTHi_G4PU1VQpBTsAXSyM&hl=pl&sa=X&ved=0ahUKEwjA75rnlYTLAhWB83IKHbLvAQoQ6AEIGzAA#v=onepage&q=measure%20virtual%20machine%20temperature%20load&f=false
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3967661/
Dear Wolk,
I have already looked at the second one "Prediction Based Proactive Thermal Virtual Machine Scheduling in Green Clouds" and it is not exactly what I am looking for. They simply place the VM in the server that has the biggest difference between CPU threshold temperature and current temperature in the hope that the scheduled VM will not exceed threshold temperature. I will however go through the first one and see if it will help.
What you are looking for is a temperature prediction strategy based on the type of the VM. There are already established models for temperature prediction (see the first link). In order to accurately predict the temperature increase, you should have some information about the VM, such as the utilization level, job size. Given these information, you can make some estimations about the future temperature.
http://faculty.cse.tamu.edu/ejkim/HPC_WEB/docs/dac08_yeo.pdf
- Badges
- Science method