[PRL] Possibly interesting talk at MIT

Mitchell Wand wand at ccs.neu.edu
Wed Nov 19 13:18:59 EST 2008


From: CSAIL Event Calendar <eventcalendar at csail.mit.edu>
To: seminars at csail.mit.edu
Date: Wed, 19 Nov 2008 11:26:28 -0500
Subject: TALK:Tuesday 11-25-08 Berkeley Orders of Magnitude

Berkeley Orders of Magnitude
Speaker: Tyson Condie
Speaker Affiliation: UC-Berkeley
Host: Sam Madden
Host Affiliation: CSAIL

Date: 11-25-2008
Time: 4:00 PM - 5:00 PM
Refreshments: 3:45 PM
Location: 32-D463 (Star)

Abstract:
In this talk I will describe our current work at Berkeley in the BOOM
(Berkeley Orders of Magnitude) project. The primary goal of BOOM is to
express the complex logic of distributed system management in a declarative
language. Building systems in a high-level declarative language enables
extensibility, which I will demonstrate by describing our design and
implementation of Evita Raced, an extensible compiler for the OverLog
language used in our declarative networking system, P2. Evita Raced is a
metacompiler -- an OverLog compiler written in OverLog. I will describe the
minimalist architecture of Evita Raced, including its extensibility
interfaces and its reuse of P2's data model and runtime engine. I will then
demonstrate the effectiveness of using a declarative language like OverLog
in expressing traditional and novel query optimizations as well as other
query manipulations in a compact and natural fashion. In the last part of
this talk, I will discuss our approach to building a set of cloud-computing
services in OverLog, including the MapReduce (Hadoop), GFS (Hadoop DFS), and
Chubby (ZooKeeper) services. I will focus on the MapReduce portion by
describing our declarative implementation of the Hadoop JobTracker, which
schedules MapReduce tasks on a cluster computing environment.
Tyson Condie is a Ph.D. candidate at UC-Berkeley.

Relevant URL(S):
For more information please contact: Sheila Marian, 617-253-1996,
sheila at csail.mit.edu
-------------- next part --------------
HTML attachment scrubbed and removed


More information about the PRL mailing list