Presentation is loading. Please wait.

Presentation is loading. Please wait.

A Framework for Elastic Execution of Existing MPI Programs Aarthi Raveendran Tekin Bicer Gagan Agrawal 1.

Similar presentations


Presentation on theme: "A Framework for Elastic Execution of Existing MPI Programs Aarthi Raveendran Tekin Bicer Gagan Agrawal 1."— Presentation transcript:

1 A Framework for Elastic Execution of Existing MPI Programs Aarthi Raveendran Tekin Bicer Gagan Agrawal 1

2 Motivation  Emergence of Cloud Computing Including for HPC Applications  Key Advantages of Cloud Computing Elasticity (dynamically acquire resources) Pay-as-you model Can be exploited to meet cost and/or time constraints  Existing HPC Applications MPI-based, use fixed number of nodes  Need to make Existing MPI Applications Elastic A Framework for Elastic Execution of Existing MPI Programs 2

3 Detailed Research Objective  To make MPI applications elastic Exploit key advantage of Cloud Computing Meet user defined time and/or cost constraints Avoid new programming model or significant recoding  Design a framework for Decision making When to expand or contract Actual Support for Elasticity Allocation, Data Redistribution, Restart A Framework for Elastic Execution of Existing MPI Programs 3

4 Outline  Research Objective  Framework Design  Run time support modules  Experimental Platform: Amazon Cloud Services  Applications and Experimental Evaluation  Conclusion A Framework for Elastic Execution of Existing MPI Programs 4

5 Framework components A Framework for Elastic Execution of Existing MPI Programs 5

6 Framework design – Approach and Assumptions  Target – Iterative HPC Applications  Assumption : Uniform work done at every iteration  Monitoring at the start of every few iterations of the time-step loop  Checkpointing and re- distribution  Calculate required iteration time based on user input A Framework for Elastic Execution of Existing MPI Programs 6

7 Framework design - Modification to Source Code  Progress checked based on current average iteration time  Decision made to stop and restart if necessary  Reallocation should not be done too frequently  If restarting is not necessary, the application continues running A Framework for Elastic Execution of Existing MPI Programs 7

8 8 Framework Design Execution flow

9 Other Runtime Steps  Steps taken to perform scaling to a different number of nodes:  Live variables and arrays need to be collected at the master node and redistributed  Read only need not be restored – just retrieve  Application is restarted with each node reading the local portions of the redistributed data. A Framework for Elastic Execution of Existing MPI Programs 9

10 Runtime support modules Decision layer  Interaction with user and application program  Constraints- Time or cost  Monitoring the progress and making a decision  Current work :  Measuring communication overhead and estimating scalability  Moving to large – type instances if necessary A Framework for Elastic Execution of Existing MPI Programs 10

11 Framework design – Modification to Source Code A Framework for Elastic Execution of Existing MPI Programs 11

12 Background – Amazon cloud  Services used in our framework :  Amazon Elastic compute cloud (EC2)  Virtual images called instances  Small instances : 1.7 GB of memory, 1 EC2 Compute Unit, 160 GB of local instance storage, 32-bit platform  Large instances : 7.5 GB of memory, 4 EC2 Compute Units, 850 GB of local instance storage, 64-bit platform  On demand, reserved, spot instances A Framework for Elastic Execution of Existing MPI Programs 12

13 Background – Amazon cloud  Amazon Simple Storage Service (S3)  Provides key - value store  Data stored in files  Each file restricted to 5 GB  Unlimited number of files A Framework for Elastic Execution of Existing MPI Programs 13

14 Runtime support modules Resource allocator  Elastic execution  Input taken from the decision layer on the number of resources  Allocating de- allocating resources in AWS environment  MPI configuration for these instances  Setting up of the MPI cluster  Configuring for password less login among nodes A Framework for Elastic Execution of Existing MPI Programs 14

15 Runtime support modules Check pointing and redistribution  Multiple design options feasible with the support available on AWS  Amazon S3  Unmodified Arrays  Quick access from EC2 instances  Arrays stored in small sized chunks  Remote file copy  Modified arrays (live arrays)  File writes and reads A Framework for Elastic Execution of Existing MPI Programs 15

16 Runtime support modules Check pointing and redistribution  Current design  Knowledge of division of the original dataset necessary  Aggregation and redistribution done centrally on a single node  Future work  Source to source transformation tool  Decentralized array distribution schemes A Framework for Elastic Execution of Existing MPI Programs 16

17 Experiments  Framework and approach evaluated using  Jacobi  Conjugate Gradient (CG )  MPICH 2 used  4, 8 and 16 small instances used for processing the data  Observation made with and without scaling the resources - Overheads 5-10%, which is negligible A Framework for Elastic Execution of Existing MPI Programs 17

18 Experiments – Jacobi A Framework for Elastic Execution of Existing MPI Programs 18

19 Experiments – Jacobi A Framework for Elastic Execution of Existing MPI Programs 19

20 Experiments – Jacobi  Matrix updated at every iteration  Updated matrix collected and redistributed at node change  Worst case total redistribution overhead – less than 2%  Scalable application – performance increases with number of nodes A Framework for Elastic Execution of Existing MPI Programs 20

21 Experiments - CG A Framework for Elastic Execution of Existing MPI Programs 21

22 Experiments - CG A Framework for Elastic Execution of Existing MPI Programs 22

23 Experiments - CG  Single vector which needs to be redistributed  Communication intensive application  Not scalable  Overheads are still low A Framework for Elastic Execution of Existing MPI Programs 23

24 Conclusion  An overall approach to make MPI applications elastic and adaptable  An automated framework for deciding the number of instances for execution  Framework tested using 2 MPI applications showing low overheads during elastic execution A Framework for Elastic Execution of Existing MPI Programs 24


Download ppt "A Framework for Elastic Execution of Existing MPI Programs Aarthi Raveendran Tekin Bicer Gagan Agrawal 1."

Similar presentations


Ads by Google