Personal tools
You are here: Home Events ICSA COLLOQUIUM TALK


— filed under:

Talk and presentation by Thomas Wenisch from University of Michigan. Title: Efficiency Challenges in Warehouse-Scale Computers.

  • Colloquium Series
When Mar 02, 2012
from 02:30 PM to 03:30 PM
Where Informatics Forum 4.31/4.33
Add event to calendar vCal


Architects and circuit designers have made enormous strides in managing the energy efficiency and peak power demands of processors and other silicon systems.  Sophisticated power management features and modes are now myriad across system components, from DRAM to processors to disks. And yet, despite these advances, typical data centers today suffer embarrassing energy inefficiencies: it is not unusual for less than 20% of a data center's multi-megawatt total power draw to flow to computer systems actively performing useful work.   Managing power and energy is challenging because individual systems and entire facilities are conservatively provisioned for rare utilization peaks, which leads to energy waste in underutilized systems and over-provisioning of physical infrastructure. Power management is particularly challenging for Online Data Intensive (OLDI) services---workloads like social networking, web search, ad serving, and machine translation that perform significant computing over massive data sets for each user request but require responsiveness in sub-second time scales.  These inefficiencies lead to worldwide energy waste measured in billions of dollars and tens of millions of metric tons of CO2.

In this talk, I discuss what, if anything, can be done to make OLDI systems more energy-proportional. Specifically, through a case study of Google's Web Search application, I will discuss the applicability of existing and proposed active and idle low-power modes to reduce the  power consumed by the primary server components (processor, memory, and disk), while maintaining tight response time constraints, particularly on 95th-percentile latency. Then, I will briefly discuss our work on PowerRouting, a proposal to dynamically switch servers among redundant power feeds to reduce overprovisioning in data center power delivery infrastructure.  

Finally, I will close with comments on our ongoing work examining server and cluster-level performance optimization for memcached, a  distributed main-memory key-value store that acts as a cache for more expensive durable stores (e.g., a DBMS) and is a key piece of  infrastructure for OLDI services.



Thomas Wenisch is the Morris Wellman Faculty Development Assistant Professor of Computer Science and Engineering at the University of  Michigan, specializing in computer architecture. Tom's prior research includes memory streaming for commercial server applications, store-wait-free multiprocessor memory systems, memory disaggregation, and rigorous sampling-based performance evaluation methodologies. His ongoing work focuses on data center architecture, energy-efficient server design, and multi-core / multiprocessor memory systems. Tom received an NSF CAREER award in 2009. Prior to his academic career, Tom was a software developer at American  Power Conversion, where he worked on data center thermal topology estimation. He is co-inventor on four patents. Tom received his Ph.D.  in Electrical and Computer Engineering from Carnegie Mellon University.

Document Actions