CS 7934 — Computer Systems Seminar, Fall 2011

Fridays, 2:00–3:00 PM, 3485 MEB

Organizer: Eric Eide


The fall 2011 offering of CS 7934 will cover a variety of systems topics, with an eye toward three goals.

The first is to increase participants' understanding of datacenter architectures and issues. Datacenters today are at the heart of large organizations—and like a heart, if the datacenter stops, the organization stops, too. Datacenter design and operation covers a myriad of issues including physical layout, power consumption and “green computing,” network design, physical and computing security, and the operating systems, middleware, and applications that run on the facility. To better understand these issues, we will study recent publications that describe practices, trends, research, and case studies related to datacenter design and operation. Particular attention will be given to cloud- and security-related topics.

The second is to be a venue for student presentations. Every student participating in the seminar will be required to give at least one “formal” research presentation during the semester. Ideally these will be presentations of students' current work, but other topics are also possible.

The third is to stay abreast of papers from recent and imminent top-tier systems conferences: e.g., SOSP, OSDI, NSDI, SIGCOMM, FAST, systems-related security conferences, and so on. Papers will be selected for their relevance to participants' research or upcoming Utah visitors.

CS 7934 is often called “the CSL seminar.” The name CSL is historic.

Mailing list

To get on the class mailing list, use Mailman to subscribe to csl-sem.


Those taking the course for credit must read all of the papers, submit a short summary of each paper prior to class (PDF, Postscript, LaTeX), participate in each discussion, and make at least one research presentation.

A summary must consist of a student's original work. Unattributed copying of material from the papers being discussed or other sources is plagiarism as will be treated as such.

We urge students to sign up for one credit if you're going to be attending anyway. Although the course is listed as “variable credit,” the course is only available for one (1) credit in all but the most exceptional circumstances. If you want to take the course for more than one credit, you must request and receive approval from the instructor.


(You can check out what we did last semester here.)

Week Date Topic(s) Facilitator Paper(s)
1 8/26 Eide no meeting — organizational email
2 9/2 datacenter overview Eide The Datacenter as a Computer: An Introduction to the Design of Warehouse-Scale Machines. Luiz André Barroso and Urs Hölzle. Synthesis Lectures on Computer Architecture #6. Morgan & Claypool. 2009.
3 9/9 low latency Burtsev It's Time for Low Latency. Stephen M. Rumble et al. In HotOS '11, May 2011.

Megastore: Providing Scalable, Highly Available Storage for Interactive Services. Jason Baker et al. In CIDR '11, Jan. 2011.

Supplementary: Google Megastore: The Data Engine Behind GAE. James Hamilton. Blog post, Jan. 2011.

Supplementary: Warehouse-Scale Computing: Entering the Teenage Decade. Luiz Andre Barrosso. Plenary talk at ICSA '11, Jun. 2011.
4 9/16 hypervisor security Chung Hwan Kim Breaking Up is Hard to Do: Security and Functionality in a Commodity Hypervisor. Patrick Colp et al. In SOSP '11, Oct. 2011. To appear.

Eliminating the Hypervisor Attack Surface for a More Secure Cloud. Jakub Szefer et al. In CCS '11, Oct. 2011. To appear.
5 9/23 cloud utility Duerig To Cloud Or Not To Cloud? Musings On Costs and Viability. Yao Chen and Radu Sion. In SOCC '11, Oct. 2011. To appear. (This is the author's copy, not the final version.)

ExPERT: Pareto-Efficient Task Replication on Grids and Clouds. Orna Agmon Ben-Yehuda et al. Technical report CS-2011-03, Technion Computer Science Department, Apr. 2011.
6 9/30 datacenter networks Wong VL2: A Scalable and Flexible Data Center Network. Albert Greenberg et al. In SIGCOMM '09, Aug. 2009.

BCube: A High Performance, Server-centric Network Architecture for Modular Data Centers. Chuanxiong Guo et al. In SIGCOMM '09, Aug. 2009.

Supplementary: DCell: A Scalable and Fault-Tolerant Network Structure for Data Centers. Chuanxiong Guo et al. In SIGCOMM '08, Aug. 2008.
7 10/7 abstractions for GPUs Sun PTask: Operating System Abstractions To Manage GPUs as Compute Devices. Christopher J. Rossbach et al. In SOSP '11, Oct. 2011. To appear.

Automatic CPU-GPU Communication Management and Optimization. Thomas B. Jablin et al. In PLDI '11, Jun. 2011.
8 10/14 no meeting — University fall break
9 10/21 potpourri Ricci PipeCloud: Using Causality to Overcome Speed-of-Light Delays in Cloud-Based Disaster Recovery. Timothy Wood et al. In SOCC '11, Oct. 2011. To appear. (This is the author's copy.)

Wresting Control from BGP: Scalable Fine-grained Route Control. Patrick Verkaik et al. In USENIX ATC '07, Jun. 2007.
10 10/28 high-speed signature matching Sharma SplitScreen: Enabling Efficient, Distributed Malware Detection. Sang Kil Cha et al. In NSDI '10, Apr. 2010.

Gnort: High Performance Network Intrusion Detection Using Graphics Processors. Giorgos Vasiliadis et al. In RAID '08, Sep. 2008.
11 11/4 disk failures Naik Failure Trends in a Large Disk Drive Population. Eduardo Pinheiro et al. In FAST '07, Feb. 2007.

An Analysis of Data Corruption in the Storage Stack. Lakshmi N. Bairavasundaram et al. In FAST '08, Feb. 2008.

Supplementary: An Analysis of Latent Sector Errors in Disk Drives. Lakshmi N. Bairavasundaram et al. In SIGMETRICS '07, Jun. 2007.
12 11/11 green storage Doowon Kim Massive Arrays of Idle Disks For Storage Archives. Dennis Colarelli and Dirk Grunwald. In SC '02, Nov. 2002.

Hibernator: Helping Disk Arrays Sleep Through the Winter. Qingbo Zhu et al. In SOSP '05, Oct. 2005.
13 11/18 datacenter programming Raju BOOM Analytics: Exploring Data-Centric, Declarative Programming for the Cloud. Peter Alvaro et al. In EuroSys '10, Apr. 2010.

Nectar: Automatic Management of Data and Computation in Datacenters. Pradeep Kumar Gunda et al. In OSDI '10, Oct. 2010.
14 11/25 no meeting — Thanksgiving break
15 12/2 the underground economy Johnson Click Trajectories: End-to-End Analysis of the Spam Value Chain. Kirill Levchenko et al. In IEEE S&P '11, May 2011.

Re: CAPTCHAs – Understanding CAPTCHA-Solving in an Economic Context. Marti Motoyama et al. In USENIX Security '10, Aug. 2010.

Supplementary: No Plan Survives Contact: Experience with Cybercrime Measurement. Chris Kanich et al. In CSET '11, Aug. 2011.
16 12/9 datacenter networking Jimei Data Center TCP (DCTCP). Mohammad Alizadeh et al. In SIGCOMM '10, Aug.–Sep. 2010.

NOX: Towards an Operating System for Networks. Natasha Gude et al. ACM SIGCOMM CCR, 38(3):105–110, Jul. 2008.

Supplementary: Applying NOX to the Datacenter. Arsalan Tavakoli et al. In HotNets '09, Oct. 2009.

Potential Papers

Upcoming and recent conference proceedings are good sources of papers for discussion. Below are links to some relevant conference series.