automated debugging in data intensive scalable computing
play

Automated Debugging In Data Intensive Scalable Computing Systems - PowerPoint PPT Presentation

Jun 21, 2023 •131 likes •397 views

Automated Debugging In Data Intensive Scalable Computing Systems Muhammad Ali Gulzar 1 , Matteo Interlandi 3 , Xueyuan Han 2 , Mingda Li 1 , Tyson Condie 1 , and Miryung Kim 1 1 University of California, Los Angeles 2 Harvard University 3

  1. Automated Debugging In Data Intensive Scalable Computing Systems Muhammad Ali Gulzar 1 , Matteo Interlandi 3 , Xueyuan Han 2 , Mingda Li 1 , Tyson Condie 1 , and Miryung Kim 1 1 University of California, Los Angeles 2 Harvard University 3 Mircrosoft 1

  2. Big Data Debugging in the Dark Develop locally Hope it works Run in cloud Bug! 1 2 3 4 Guesswork 5 Map Reduce 2

  3. Motivating Example • Alice writes a Spark program that identifies, for each state in the US, the delta between the minimum and the maximum snowfall reading for each day of any year and for any particular year . Zip Code Date Snow Fall 99504 01/01/1994 245mm 99504 01/01/1993 85mm 90031 02/01/1991 0mm … … … 3

  4. Problem Definition • Using a test function, a user can specify incorrect results TextFile FlatMap GroupByKey Map Output AK , 01/01 , 21251 AK, 01/01 , 304.8 AK , 01/01 , [304.8, 21336, 245, 85] 99504, 01/01/1992 , 1ft AK , 03/01, 114.5 AK, 1992 , 304.8 AK , 03/01 , [30.5 , 145] 99504, 03/01/1992 , 0.1ft AK , 1992 , 274.3 AK, 03/01 , 30.5 AK , 1992 , [304.8 , 30.5] 99504, 01/01/1993 , 70in AK , 1993 , 21251 AK, 1992 , 30.5 AK , 1993 , [21336, 145, 85] 99504, 03/01/1993 , 145mm AK , 1994 , 0 AK, 01/01 , 21336 AK , 1994 , [245] def test(key:String, delta: Float) : Boolean = { 99504, 01/01/1994 , 245mm CA , 02/01, 0 AK, 1993 , 21336 CA , 02/01 , [0] delta < 6000 99504, 01/01/1993 , 85mm CA , 1991 , 0 AK, 03/01 , 145 CA , 1991 , [0] } 90031, 02/01/1991 , 0mm AK, 1993 , 145 AK, 01/01 , 245 AK, 1994 , 245 …. …. Given a test function, the goal is to identify a minimum subset of the input that is able to reproduce the same test failure. 4

  5. Existing Approach 1: Data Provenance for Spark TextFile FlatMap GroupByKey Map Output AK , 01/01 , 21251 AK, 01/01 , 304.8 AK , 01/01 , [304.8, 21336, 245, 85] 99504, 01/01/1992 , 1ft AK , 03/01, 114.5 AK, 1992 , 304.8 AK , 03/01 , [30.5 , 145] 99504, 03/01/1992 , 0.1ft AK , 1992 , 274.3 AK, 03/01 , 30.5 AK , 1992 , [304.8 , 30.5] 99504, 01/01/1993 , 70in AK , 1993 , 21251 AK, 1992 , 30.5 AK , 1993 , [21336, 145, 85] 99504, 03/01/1993 , 145mm AK , 1994 , 0 AK, 01/01 , 21336 AK , 1994 , [245] 99504, 01/01/1994 , 245mm CA , 02/01, 0 AK, 1993 , 21336 CA , 02/01 , [0] 99504, 01/01/1993 , 85mm CA , 1991 , 0 AK, 03/01 , 145 CA , 1991 , [0] 90031, 02/01/1991 , 0mm AK, 1993 , 145 AK, 01/01 , 245 AK, 1994 , 245 …. …. It over-approximates the scope of failure-inducing inputs i.e. records in the faulty key-group are all marked as faulty 5

  6. Existing Approach 2: Delta Debugging • Delta Debugging performs a systematic binary search-like procedure on the input dataset using a test oracle function TextFile FlatMap GroupByKey Map Output AK ,01/01 , 304.8 AK ,1992 , 304.8 AK ,03/01 , 30.5 99504, 01/01/1992 , 1ft AK ,1992 , 30.5 AK , 01/01 , 21251 AK ,01/01 , [304.8, 21336, 245, 85] 1 99504, 03/01/1992 , 0.1ft AK ,01/01 , 21336 AK , 03/01 , 114.5 AK ,03/01 , [30.5 , 145] 99504, 01/01/1993 , 70in AK ,1993 , 21336 AK , 1992 , 274.3 AK ,1992 , [304.8 , 30.5] 99504, 03/01/1993 , 145mm AK ,03/01 , 145 AK , 1993 , 21251 AK ,1993 , [21336, 145, 85] 99504, 01/01/1994, 245mm AK ,1993 , 145 AK , 1994 , 0 AK ,1994 , [245] 2 99504, 01/01/1993, 85mm AK ,01/01 , 245 CA , 02/01 , 0 CA ,02/01 , [0] 90031, 02/01/1991, 0mm AK ,1994 , 245 CA , 1991 , 0 CA ,1991 , [0] …. …. It does not prune input records known to be irrelevant because of the lack of semantic understanding of data-flow operators 6

  7. Existing Approach 2: Delta Debugging • Delta Debugging performs a systematic binary-like search on the input dataset using a test oracle function TextFile FlatMap GroupByKey Map Output AK ,01/01 , 304.8 AK , 01/01 , 21031 AK ,01/01 , [304.8, 21336] 99504, 01/01/1992 , 1ft AK ,1992 , 304.8 AK , 03/01 , 0 1 AK ,03/01 , [30.5] 99504, 03/01/1992 , 0.1ft AK ,03/01 , 30.5 AK , 1992 , 274.3 AK ,1992 , [304.8 , 30.5] 99504, 01/01/1993 , 70in 2 AK ,1992 , 30.5 AK , 1993 , 0 AK ,1993 , [21336] AK ,01/01 , 21336 AK ,1993 , 21336 Run 2 It does not prune input records known to be irrelevant because of the lack of semantic understanding of data-flow operators 7

  8. Existing Approach 2: Delta Debugging • Delta Debugging performs a systematic binary-like search on the input dataset using a test oracle function TextFile FlatMap GroupByKey Map Output AK ,01/01 , 304.8 AK , 01/01 , 0 AK ,01/01 , [304.8] 99504, 01/01/1992 , 1ft AK ,1992 , 304.8 AK , 03/01 , 0 AK ,03/01 , [30.5] 99504, 03/01/1992 , 0.1ft AK ,03/01 , 30.5 AK , 1992 , 274.3 AK ,1992 , [304.8 , 30.5] 99504, 01/01/1993 , 70in AK ,1992 , 30.5 Run 3 It does not prune input records known to be irrelevant because of the lack of semantic understanding of data-flow operators 8

  9. Existing Approach 2: Delta Debugging • Delta Debugging performs a systematic binary-like search on the input dataset using a test oracle function TextFile FlatMap GroupByKey Map Output 99504, 01/01/1992 , 1ft AK , 01/01 , 0 AK ,01/01 , 21336 AK ,01/01 , [21336] 99504, 03/01/1992 , 0.1ft AK , 1993 , 0 AK ,1993 , 21336 AK ,1993 , [21336] 99504, 01/01/1993 , 70in Run 4 It does not prune input records known to be irrelevant because of the lack of semantic understanding of data-flow operators 9

  10. Existing Approach 2: Delta Debugging • Delta Debugging performs a systematic binary-like search on the input dataset using a test oracle function TextFile FlatMap GroupByKey Map Output 99504, 01/01/1992 , 1ft AK , 01/01 , 0 AK ,01/01 , 304.8 AK ,01/01 , [304.8] 99504, 03/01/1992 , 0.1ft AK , 1992 , 0 AK ,1992 , 304.8 AK ,1992 , [304.8] 99504, 01/01/1993 , 70in Run 5 It does not prune input records known to be irrelevant because of the lack of semantic understanding of data-flow operators 10

  11. Existing Approach 2: Delta Debugging • Delta Debugging performs a systematic binary-like search on the input dataset using a test oracle function TextFile FlatMap GroupByKey Map Output 99504, 01/01/1992 , 1ft AK , 03/01 , 0 AK ,03/01 , 30.5 AK ,03/01 , [30.5] 99504, 03/01/1992 , 0.1ft AK , 1992 , 0 AK ,1992 , 30.5 AK ,1992 , [30.5] 99504, 01/01/1993 , 70in Run 6 It does not prune input records known to be irrelevant because of the lack of semantic understanding of data-flow operators 11

  12. Existing Approach 2: Delta Debugging • Delta Debugging performs a systematic binary-like search on the input dataset using a test oracle function TextFile FlatMap GroupByKey Map Output 99504, 01/01/1992 , 1ft AK , 01/01 , 0 AK ,01/01 , 21336 AK ,01/01 , [21336] 99504, 03/01/1992 , 0.1ft AK , 1993 , 0 AK ,1993 , 21336 AK ,1993 , [21336] 99504, 01/01/1993 , 70in Run 7 It does not prune input records known to be irrelevant because of the lack of semantic understanding of data-flow operators 12

  13. Existing Approach 2: Delta Debugging • Delta Debugging performs a systematic binary-like search on the input dataset using a test oracle function TextFile FlatMap GroupByKey Map Output AK , 01/01 , 0 AK ,03/01 , 30.5 AK ,01/01 , [21336] 99504, 01/01/1992 , 1ft AK , 03/01 , 0 AK ,1992 , 30.5 AK ,03/01 , [30.5] 99504, 03/01/1992 , 0.1ft AK , 1992 , 0 AK ,01/01 , 21336 AK ,1992 , [30.5] 99504, 01/01/1993 , 70in AK , 1993 , 0 AK ,1993 , [21336] AK ,1993 , 21336 Run 8 It does not prune input records known to be irrelevant because of the lack of semantic understanding of data-flow operators 13

  14. Existing Approach 2: Delta Debugging • Delta Debugging performs a systematic binary-like search on the input dataset using a test oracle function TextFile FlatMap GroupByKey Map Output AK ,01/01 , 304.8 AK , 01/01 , 21031 AK ,01/01 , [304.8, 21336] 99504, 01/01/1992 , 1ft AK ,1992 , 304.8 AK , 1992 , 0 AK ,1992 , [304.8] 99504, 03/01/1992 , 0.1ft AK ,01/01 , 21336 AK , 1993 , 0 AK ,1993 , [21336] 99504, 01/01/1993 , 70in AK ,1993 , 21336 Run 9 It does not prune input records known to be irrelevant because of the lack of semantic understanding of data-flow operators 14

  15. Automated Debugging in DISC with BigSift Output: Minimum Fault-Inducing Input: A Spark Program, A Test Function Input Records Data Provenance + Delta Debugging Prioritizing Bitmap based Test Predicate Backward Test Pushdown Traces Memoization 15

  16. Optimization 1: Test Predicate Pushdown • Observation: During backward tracing, data provenance traces through all partitions even though only a few partitions contain faulty intermediate data. Test Test Test Test Test Test Test Without Test Pushdown With Test Pushdown If applicable, BigSift pushes down the test function to test the output of combiners in order to isolate the faulty partitions. 16

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend

Automated Debugging In Data Intensive Scalable Computing Systems Muhammad Ali Gulzar 1 Siman Wang

Automated Debugging In Data Intensive Scalable Computing Systems Muhammad Ali Gulzar 1 Siman Wang

Automated Debugging In Data Intensive Scalable Computing Systems Muhammad Ali Gulzar 1 Siman Wang 1,2 Miryung Kim 1 1 University of California, Los Angeles 2 Hunan University 1 Big Data Debugging in the Dark Develop locally Hope it works Run

392 views • 22 slides
for Data Intensive Scalable Computing CAP3 Gene Assembly Program Compute intensive

for Data Intensive Scalable Computing CAP3 Gene Assembly Program Compute intensive

Architecture and Performance of Runtime Environments for Data Intensive Scalable Computing CAP3 Gene Assembly Program Compute intensive application Embarrassingly parallel operation All runtimes performs equally well Measured

150 views • 3 slides
MapReduce Data Intensive Computing Data-intensive computing is a class of parallel

MapReduce Data Intensive Computing Data-intensive computing is a class of parallel

MapReduce Data Intensive Computing Data-intensive computing is a class of parallel computing applications which use a data parallel approach to processing large volumes of data typically terabytes or petabytes in size and typically

1.01k views • 51 slides
Debugging Scalable Applications on the XT May 2nd 2009 Chris Gottbrath Director, Product

Debugging Scalable Applications on the XT May 2nd 2009 Chris Gottbrath Director, Product

Debugging Scalable Applications on the XT May 2nd 2009 Chris Gottbrath Director, Product Management Debugging Scalable Applications Intro Challenges Products Scalability Interactive Subset Debugging Batch

727 views • 30 slides
Limitlessly Scalable Storage for Capacity-Intensive Computing Meet Cloudian S3-compatible

Limitlessly Scalable Storage for Capacity-Intensive Computing Meet Cloudian S3-compatible

Limitlessly Scalable Storage for Capacity-Intensive Computing Meet Cloudian S3-compatible object storage Appliances and Software-defined Storage Start small, scale to over an exabyte Hybrid cloud and multi-cloud ready 2 See

257 views • 7 slides
Scalable Post-Mortem Debugging Abel Mathew CEO - Backtrace amathew@backtrace.io @nullisnt0

Scalable Post-Mortem Debugging Abel Mathew CEO - Backtrace amathew@backtrace.io @nullisnt0

Scalable Post-Mortem Debugging Abel Mathew CEO - Backtrace amathew@backtrace.io @nullisnt0 Debugging or Sleeping? Debugging Debugging: examining (program state, design, code, output) to identify and remove errors from software.

464 views • 31 slides
What is Advanced Research Computing? Data Supercomputing Computationally Mining Intensive

What is Advanced Research Computing? Data Supercomputing Computationally Mining Intensive

Accelerate Discovery with ACENET www.acceleratediscovery.ca What is Advanced Research Computing? Data Supercomputing Computationally Mining Intensive High Parallel Performance Big Data Storage Computing Computing (HPC) High

399 views • 4 slides
Data-Intensive Distributed Computing 451/651 (Fall 2020) Part 1: Introduction to Big Data Ali

Data-Intensive Distributed Computing 451/651 (Fall 2020) Part 1: Introduction to Big Data Ali

Data-Intensive Distributed Computing 451/651 (Fall 2020) Part 1: Introduction to Big Data Ali Abedi These slides are available at https://www.student.cs.uwaterloo.ca/~cs451/ 1 1 2 3 Big Data 4 Lets see what big data is and where it

744 views • 47 slides
Data-Intensive Distributed Computing 431/631 (Fall 2020) Part 1: Introduction to Big Data Ali

Data-Intensive Distributed Computing 431/631 (Fall 2020) Part 1: Introduction to Big Data Ali

Data-Intensive Distributed Computing 431/631 (Fall 2020) Part 1: Introduction to Big Data Ali Abedi These slides are available at https://www.student.cs.uwaterloo.ca/~cs451/ 1 1 2 3 Big Data 4 Lets see what big data is and where it

778 views • 44 slides
Data-Intensive Distributed Computing CS 431/631 451/651 (Fall 2020) Part 7: Data Mining (2/4)

Data-Intensive Distributed Computing CS 431/631 451/651 (Fall 2020) Part 7: Data Mining (2/4)

Data-Intensive Distributed Computing CS 431/631 451/651 (Fall 2020) Part 7: Data Mining (2/4) Ali Abedi These slides are available at https://www.student.cs.uwaterloo.ca/~cs451 This work is licensed under a Creative Commons

929 views • 51 slides
Data-Intensive Distributed Computing CS 431/631 451/651 (Fall 2019) Part 6: Data Mining (3/4)

Data-Intensive Distributed Computing CS 431/631 451/651 (Fall 2019) Part 6: Data Mining (3/4)

Data-Intensive Distributed Computing CS 431/631 451/651 (Fall 2019) Part 6: Data Mining (3/4) November 5, 2019 Ali Abedi Thanks to Jure Leskovec, Anand Rajaraman, Jeff Ullman (Stanford University) These slides are available at

757 views • 52 slides
Data-Intensive Distributed Computing CS 431/631 451/651 (Winter 2019) Part 6: Data Mining (1/4)

Data-Intensive Distributed Computing CS 431/631 451/651 (Winter 2019) Part 6: Data Mining (1/4)

Data-Intensive Distributed Computing CS 431/631 451/651 (Winter 2019) Part 6: Data Mining (1/4) October 29, 2019 Ali Abedi These slides are available at https://www.student.cs.uwaterloo.ca/~cs451 This work is licensed under a Creative Commons

842 views • 35 slides
Data-Intensive Distributed Computing 431/451/631/651 (Fall 2020) Part 1: MapReduce Algorithm

Data-Intensive Distributed Computing 431/451/631/651 (Fall 2020) Part 1: MapReduce Algorithm

Data-Intensive Distributed Computing 431/451/631/651 (Fall 2020) Part 1: MapReduce Algorithm Design (1/3) Ali Abedi These slides are available at https://www.student.cs.uwaterloo.ca/~cs451/ 1 Agenda for today Abstraction Storage/computing

690 views • 37 slides
Data-Intensive Distributed Computing CS 431/631 451/651 (Fall 2020) Part 9: Real-Time Data

Data-Intensive Distributed Computing CS 431/631 451/651 (Fall 2020) Part 9: Real-Time Data

Data-Intensive Distributed Computing CS 431/631 451/651 (Fall 2020) Part 9: Real-Time Data Analytics (1/2) Ali Abedi These slides are available at https://www.student.cs.uwaterloo.ca/~cs451 This work is licensed under a Creative Commons

757 views • 36 slides
Automated Debugging for Arbitrarily Long Executions Cristian Zamfir, Baris Kasikci, Johannes

Automated Debugging for Arbitrarily Long Executions Cristian Zamfir, Baris Kasikci, Johannes

Automated Debugging for Arbitrarily Long Executions Cristian Zamfir, Baris Kasikci, Johannes Kinder, Edouard Bugnion, George Candea Debugging is Hard Debugging = diagnose + fix the root cause May take days-months to diagnose bugs in the

945 views • 62 slides
Extreme Data-Intensive Scientific Computing Alex Szalay JHU Big Data in Science Data

Extreme Data-Intensive Scientific Computing Alex Szalay JHU Big Data in Science Data

Extreme Data-Intensive Scientific Computing Alex Szalay JHU Big Data in Science Data growing exponentially, in all science All science is becoming data-driven This is happening very rapidly Data becoming increasingly

347 views • 23 slides
1 University of California, Davis 2 University of Athens, Greece 3 LogicBlox Inc., Atlanta, USA UC

1 University of California, Davis 2 University of Athens, Greece 3 LogicBlox Inc., Atlanta, USA UC

Sven Khler 1 Bertram Ludscher 1,3 Yannis Smaragdakis 2,3 1 University of California, Davis 2 University of Athens, Greece 3 LogicBlox Inc., Atlanta, USA UC DAVIS Department of Computer Science Datalog2.0, Vienna Logic Weeks Outline

646 views • 38 slides
Language-integrated Provenance Stefan Fehrenbach James Cheney PPDP 2016 A database Agencies

Language-integrated Provenance Stefan Fehrenbach James Cheney PPDP 2016 A database Agencies

Language-integrated Provenance Stefan Fehrenbach James Cheney PPDP 2016 A database Agencies oid name based_in phone 1 EdinTours Edinburgh 8740 2489 123 2 Burnss Glasgow 9307 2394 104 ExternalTours oid name destination type

737 views • 39 slides
Propagation and Provenance Need to go Beyond . . . Model Fusion: We . . . of Uncertainty in

Propagation and Provenance Need to go Beyond . . . Model Fusion: We . . . of Uncertainty in

Outline Need for Uncertainty . . . Case Study: Seismic . . . Propagation and Provenance Need to go Beyond . . . Model Fusion: We . . . of Uncertainty in Optimal Stationary . . . Dynamic Sensors: . . . Cyberinfrastructure-Related Home Page

391 views • 24 slides
Privacy Issues of Privacy Issues of Provenance in Provenance in Electronic Healthcare

Privacy Issues of Privacy Issues of Provenance in Provenance in Electronic Healthcare

Tam s s Kifor Kifor, , Tam L szl szl Z. Z. Varga Varga, , L Sergio lvarez lvarez, , Sergio Javier V V zquez zquez- -Salceda Salceda, , Javier Steven Willmott Willmott Steven Privacy Issues of

470 views • 20 slides
1 2 3 Scientific workflows are similar to traditional scripts, in that they are used to automate

1 2 3 Scientific workflows are similar to traditional scripts, in that they are used to automate

1 2 3 Scientific workflows are similar to traditional scripts, in that they are used to automate computational pipelines e.g. data analysis steps.. Sci-wfs can be more scalable (e.g. may support pipeline and task parallelism etc) They are also

325 views • 28 slides
Enabling Reproducible Computing on the EPOS ICS-D Alessandro Spinuso (KNMI), Daniele Bailo (INGV),

Enabling Reproducible Computing on the EPOS ICS-D Alessandro Spinuso (KNMI), Daniele Bailo (INGV),

Enabling Reproducible Computing on the EPOS ICS-D Alessandro Spinuso (KNMI), Daniele Bailo (INGV), Matser Jonas (KNMI) Chris Card (BGS), Jean-Baptiste Roquencourt (BRGM) Wayne Shelley (BGS) Computational Earth Science (CES) European Plate

322 views • 16 slides
Provenance Collection on GENI Experimental Networks GUSH and Twister on Planetlab GUSH deploys

Provenance Collection on GENI Experimental Networks GUSH and Twister on Planetlab GUSH deploys

Provenance Collection on GENI Experimental Networks GUSH and Twister on Planetlab GUSH deploys Node 0 Head node of BFS application and application. Called Node 0 in Partition graph monitors status. PlanetLab slice Its like a shell or

510 views • 4 slides
Prayers in the Psalms Introduction Prayer book Psalter Different types of prayers

Prayers in the Psalms Introduction Prayer book Psalter Different types of prayers

Prayers in the Psalms Introduction Prayer book Psalter Different types of prayers 3 interests literary, theological, historical for interpreters of the Psalms Petitionary Psalms Psalm 3:7 Arise, O Lord! Save Me!

221 views • 7 slides

More recommend