Workshop Summary

Co-Chairs: Armando Fox ( and Ben Sigelman (

Organizing Committee:
  • Dave Patterson (Berkeley)
  • Chuck Thacker (MSR)
  • Partha Ranganathan (HP)
  • Ricardo Bianchini (Rutgers)
  • Amin Vahdat (UCSD)

When / Where:

The [half-day] workshop will be held on the morning of Sunday, June 21st at ISCA 2009 in Austin, TX. The complete workshop and tutorial schedule for ISCA 2009 can be found here.

Workshop format:

The workshop will be built around a refereed poster session and invited speakers.

This workshop should be something other than a mini-conference; specifically, we hope that a substantial poster session will give attendees time to interact and learn from each other. We hope that this format will promote discussion and foster relationships that talks alone cannot, especially since there's a great opportunity at ISCA to get those of us who primarily think about the architecture of single systems to interact with those of us who primarily think about the architecture of distributed systems and the management of large datacenters.

Workshop program:

Since this is a half-day workshop with a lot of ground to cover, the schedule is tight.

8:30am -- Settling down, opening remarks
8:40am -- Jim Larus (Microsoft Research): "Hardware can Make Data Center Software Simpler and More Robust" (slides)
9:15am -- Sarah Bird (UC Berkeley): "Fixing Performance Counters: Performance Monitoring Hardware for the Datacenter" (slides)
9:50am -- Quick break, poster session setup
9:52am -- Poster session starts as soon as possible
(10:00am --> 10:30am: official ISCA break, concurrent with poster session)
10:50am -- Richard Kaufmann (HP's Scalable Computing Infrastructure group): "Server Issues in Mainstream Clusters and Scale-out Datacenters” (slides forthcoming)
11:25am -- Chuck Thacker (Microsoft Research): "Rethinking data centers: Power, Packaging, and Networks" (slides)
12:00pm -- brief closing remarks

Call for posters:

Poster submission deadline: May 5, 2009, 11:59pm PT
Notification of acceptance: May 19th, 2009
Poster submission guidelines

The datacenter has become both an architectural building block and the backbone of modern internet applications, yet no forums exist where the cross-fertilization of knowledge from architecture and systems can be brought to bear on their design and operation. Datacenter-level power management and prediction, performance analysis, multi-tier storage systems, and improved resource isolation under virtualization are examples of areas that could benefit from active research in architecture. The first annual workshop on "Architectural Concerns in Large Datacenters" will provide such a forum through a series of invited talks and refereed posters. Poster submissions should relate to any of the following topics:
  • workload characterization for datacenters or cloud computing
  • micro-architectural and single-system concerns in datacenters:
    • virtualization and intra-machine resource isolation in the context of applications which span datacenters
    • solid-state storage for large-scale distributed systems
    • single-machine power management and prediction for datacenter workloads
    • special-purpose microprocessors or hardware acceleration for important datacenter workloads
    • hardware and/or OS support for distributed profiling and debugging
    • fault-tolerance in commodity hardware and minimizing the impact of isolated hardware failures in a datacenter environment
  • macro-architectural concerns in datacenters:
    • network design, modeling, and prediction for large-scale datacenter workloads
    • power management and prediction at the scale of racks, buildings, or facilities
    • language design for large-scale datacenter-spanning applications
    • prediction of datacenter-scale resource usage
    • resource isolation in shared services, both of per-machine resources and of the network fabric
    • models for very large-scale data processing
    • datacenter-level scheduling and multi-machine process virtualization
    • large-scale system monitoring for anomaly detection and performance analysis
    • More efficient power distribution from the grid to the POL regulators
    • More efficient cooling for single machines and for large deployments

Ben Sigelman,
Jun 24, 2009, 6:46 AM
Ben Sigelman,
Jun 24, 2009, 6:46 AM
Ben Sigelman,
Jun 24, 2009, 6:46 AM