​We describe the design and implementation of DeepDive, a system for transparently identifying and managing performance interference between virtual machines (VMs) co-located on the same physical machine in Infrastructure-as-a-Service cloud environments.

DeepDive successfully addresses several important challenges, including the lack of performance information from applications, and the large overhead of detailed interference analysis. We first show that it is possible to use easily-obtainable, low-level metrics to clearly discern when interference is occurring and what resource is causing it. Next, using realistic workloads, we show that DeepDive quickly learns about interference across co-located VMs. Finally, we show DeepDive’s ability to deal efficiently with interference when it is detected, by using a low-overhead approach to identifying a VM placement that alleviates interference.

About Dejan Novakovic

I obtained a MSc degree in school of Computer Science and Automatics from University of Novi Sad (FTN), Serbia, in 2008 where I was awarded a prestige award «Mileva Maric Einstein» for the best master thesis in the field of mathematical and informatics sciences.

Since October 2009, I am a PhD candidate under the supervision of Professor Dejan Kostic, in the Networked Systems Laboratory (NSL), EPFL.

Personal Site

El evento se impartirá en inglés