Dave Andersen

Before sending email:

• Reviews: I am currently refusing all requests to review for publications that follow the IEEE or Elsevier copyright policy.
• Potential students: Please click here for more information.

Teaching

• Spring 2019: 15-712 - Graduate Advanced Operating Systems & Distributed Systems. Previous semesters: [ Fall 2014, Fall 2013, Spring 2011, Spring 2010, Fall 2007]
• Spring 2014: 15-440 - Distributed Systems. Previous semesters: [ Fall 2012, Fall 2011, Fall 2010, Fall 2009]
• Spring 2012: 15-829: Memory and Resource Efficient Computing (special topics course)
• Spring 2009: 15-849 - Low Power Computing, a grad elective / special topics course.
• Fall 2008: 15-441 - Introduction to Computer Networks. Previous semesters: [Spring 2008] [Fall 2006] [Spring 2005]
• Spring 2007: 15-744 - Graduate Computer Networks
• Fall 2005: 15-849 - Networking in Challenging Environments

Latest Research

MemC3: Compact and Concurrent Memcache with Dumber Caching and Smarter Hashing

This paper introduces a new mechanism for memory-efficient, multi-reader concurrent hashing, called optimistic cuckoo hashing, and uses it to make Memcached scale better, run faster, and use less memory.

Stronger Semantics for Low-Latency Geo-Replicated Storage

This NSDI 2013 paper presents Eiger, a follow-on system to our earlier work on causally consistent geo replicated storage in COPS. Eiger improves upon cops by supporting Cassandra's rich data model (instead of just a basic key-value store); supporting write transactions in addition to read transactions; and fixing the problem in COPS of an unsightly buildup of state if one of the datacenters fails.

Research

My research interests are in computer systems in the networked environment. I have a particular interest in exploring the systems and data structure response to evolving hardware trends near the end of Moore's law, memory and power-efficient computing, and robust distributed systems.

Papers and Publications

My Google Scholar Profile

My reesarch group's code is mostly released on GitHub.com/efficient

Current Projects

• The Intel Center for Science and Technology (ISTC) for Visual Cloud Computing - enabling large-scale machine learning on video streams
• Using RDMA in Internet datacenters, capped by eRPC, an RPC system built for datacenters
• CuckooSwitch, libcuckoo, and the Cuckoo Filter
• FAWN (Fast Array of Wimpy Nodes) - creating a low-power cluster architecture for data-intensive workloads.

Older projects:

• AIP - the Accountable Internet Protocol
• Data-Oriented Transfer (DOT)
• The Datapository - a collaborative facility for network monitoring, analysis, and data storage
• Perspectives - using multi-path probing to improve the security of SSH and SSL authentication
• Resilient Overlay Networks (RON)
• Multi-homed Overlay Networks (MONET)
• The Congestion Manager (CM)
• Emulab

Source Code

• SILT(2011)
• XIA (2011-ongoing)
• Feed-forward Bloom Filters (2009-2011)
• FAWN-KV (2009)
• Mark-and-Sweep, a tool for wifi and broadband network connectivity scanning
• DOT (data-oriented transfer) (2005-2008)
• BGP prediction code from AIP sigcomm 2008 paper.
• MONET - Multi-Homed Overlay Networks (2005).
• RON - Resilient Overlay Networks code and data (2001-2006).
• Congestion Manager (2000).

---

Advice

I've been writing a few blog posts with bits and pieces of CS Ph.D.-relevant advice I haven't seen elsewhere:

• On the Ph.D. (to Ph.D. or not to Ph.D.?)
• Two Examples from the Computer Science Review and Publication Process
• Reflecting on CS Graduate Admissions
• Finances for CS Ph.D. students
• Finances for Recent CS Ph.D.s

Professional activities

• Program Committee ("Heavy"), OSDI 2014, OSDI 2010
• Program Committee, Symposium on Experimental Algorithms, 2014
• Member, DARPA Information Sciences and Technology (ISAT) advisory group.
• Program Committee ("Heavy"), SOSP 2013, SOSP 2009
• Program Committee, NSDI 2012, 2006
• Program Committee, HotNets 2011
• Program co-chair, NSDI 2011
• Program Committee ("Heavy"), SIGCOMM 2009, 2008
• Older PC service: Program co-chair of HotNets 2008, PC member HotNets 2007, Program co-chair, WORLDS '06 (USENIX Workshop on Real Large Distributed Systems), Internet Measurement Conference 2006 HotDep 2006 WORLDS '05, '04 USENIX 2005

About me

I completed my Ph.D. at MIT in December 2004. Prior to that, I received an MS in computer science from MIT in 2001, and BS degrees in biology and computer science from the University of Utah. In 1995, I co-founded an Internet Service Provider in Salt Lake City, Utah. My blog is online. (as is my old blog.)

Non-Research Software

• coursegen - a small set of ruby scripts/etc. to semi-auto-generate course web pages. Takes a YAML input of the list of classes, topics, etc., for the semester and outputs HTML and an iCal file. This is what generated the pages for my above-linked courses.
• paper_skel - the LaTeX template my research group uses for starting papers. Emits USENIX and ACM systems conference style-compatible PDF with less ugliness than the default ACM style file.

Random Things

• Dave's guide to Running in Pittsburgh - I have a bit of a running problem, so I put together a list of nice runs in the area.
• My list and map of running routes in Boston
• The Usable Mac - my collection of things to make my mac use a little more efficient (and bring in some unixy goodness). 2nd half is CMU specific - integrating with kerberos/AFS/etc.
• The Pi Searcher - search for arbitrary substrings in the first 200 million digits of Pi.
• Compound interest calculators of various sorts - I got sick of bloated, ad-laden sites for these that didn't do what I wanted and that didn't work well on my phone.

[Short bio] [CV] [Personal web page] [Running and Triathlon] [Climbing]