{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

lect-13-project-proposals - Lazy Evaluation for MapReduce...

Info icon This preview shows pages 1–3. Sign up to view the full content.

View Full Document Right Arrow Icon
5/20/2011 1 CSE503: SOFTWARE ENGINEERING PROJECT PROPOSALS David Notkin Spring 2011 Lazy Evaluation for MapReduce Workflows Kristi Morton, Magdalena Balazinska, Dan Grossman, and Christopher Olston Motivating Scenario Data deluge in sciences LSST workflow process 30TB of new data every day Only a subset of data needed High-latency analysis tasks in workflows On MapReduce (MR) can take hours to run Limits scientific discovery Goal: Make workflows efficient by being lazy: only run on region of interest. LSST Workflow MR Job 1 Clean data MR Job 2 Extract/transform features MR Job 3 Classify particles for a region of space (x,y,z) on some property 30 TB Lazy processing: observe usage in workflow and only processes data of interest. Any additional data is computed on demand, per user’s request.
Image of page 1

Info icon This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
5/20/2011 2 Lazy Evaluation for MapReduce Workflows Context for Project # 2: Continuation from Project 1 Workflows expressed as PigLatin scripts Use User Defined Function (UDF) for lazy evaluation framework Lazy Evaluation for MapReduce Workflows To produce the lazy materialized view, UDF needs: 1. PigLatin workflow script 2. Fields needed by user: e.g. gas, temp 3. Delimiter REGISTER udfs.jar; gas43 = LOAD ’gas43full’ USING BinStorage() AS (pid:long, mass:double, px:double, py:double, pz:double,temp:double, ...); gas = FILTER gas43 BY pid is not null AND mass is not null AND px is not null AND py is not null AND pz is not null AND temp is not null ...; regionA = FILTER gas BY temp > udfs.VirialTemp (Rvir,Mvir,pid) AND px >= -0.5 AND px < -0.25 AND py >= 0 AND py < 0.5 AND pz >= -0.5 AND pz < 0; STORE regionA INTO ’result’ USING BinStorage() Lazy Evaluation for MapReduce Workflows Gas Temp a 200 b 150 c 1000 d 999 All other fields with pointers to PigLatin scripts to generate them on the fly (i.e. the “closures” part We generate the following materialized view + metadata: Project 2 work to do Previous project generated simple metadata for on demand fields Naïve approach: each field pointed to full workload script (i.e. computed all fields) Need to have fields point to appropriate subset to compute only the data for the field not all fields Need to generate subset scripts to compute individual fields Involves looking at level of execution plan tree and pruning off subtrees Will need to hack Pig query compiler, which is a nontrivial task
Image of page 2
Image of page 3
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

What students are saying

  • Left Quote Icon

    As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

    Student Picture

    Kiran Temple University Fox School of Business ‘17, Course Hero Intern

  • Left Quote Icon

    I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

    Student Picture

    Dana University of Pennsylvania ‘17, Course Hero Intern

  • Left Quote Icon

    The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

    Student Picture

    Jill Tulane University ‘16, Course Hero Intern