Industry Internship Opportunity


Project: Public Cloud Workload Analysis
Location: Sunnyvale, CA

Job Summary

It is widely understood that public cloud workload to behave very different from the Google internal workloads. One evidence is that the lifetime of VMs outlive typical internal google jobs running in shared borg or dedicated fleets, which is usually composed of recurring instances of shorter jobs. Another example is client workloads that demand absolute performance guarantees while internal google workloads have more relaxed performance demands (e.g., perf/ total cost of ownership). The main goal of this project is to analyze the relevant logs to better understand the enterprise workload and to come up with efficient ways to schedule them. The result of the analysis will add to our understanding of the enterprise workload and our efforts to improve the efficiency of our data center resources.


An ideal candidate will be someone with a mix of traditional SWE/architecture skills as well as performance data analysis/modeling and ML background. If you have any students who might be interested, please let us know so that we can connect the resumes to the right recruiting contacts.

About the SysInfra Cloud Performance Team

The Cloud performance team within Systems Infrastructure is responsible for ensuring that all Cloud hardware platforms are designed and optimized (up and down the stack) for the applications that run on it. We model and analyze key workloads, track trends in architecture and provide input to Cloud and Platforms leadership on future platform design. Our mission also includes steering customers to the most optimal platform for their workload and tuning a platform for optimal performance. We are looking for SWEs and computer architects with a background in analytics and machine learning to help design custom systems tailored for Cloud workloads. We have a good track record of publications to Systems/Architecture conferences (e.g., NSDI/ISCA) as well as ML conferences like ICML.

Our peer teams under SysInfra Performance (formally Platform Performance) umbrella also focuses on (and is hiring FTE and interns for) analyzing and improving the performance of every system developed here for the areas of Compute, Networking, Storage, Search, and Accelerators.


Interested students should send their applications to Hui Huang or to Chelsea Llull.