7
Personal Cloud Controller (PCC) Yuan Luo 1 , Shava Smallen 2 , Beth Plale 1 , Philip Papadopoulos 2 1 School of Informatics and Computing, Indiana University Bloomington 2 San Diego Supercomputer Center, University of California San Diego

Personal Cloud Controller (PCC) Yuan Luo 1, Shava Smallen 2, Beth Plale 1, Philip Papadopoulos 2 1 School of Informatics and Computing, Indiana University

Embed Size (px)

Citation preview

Page 1: Personal Cloud Controller (PCC) Yuan Luo 1, Shava Smallen 2, Beth Plale 1, Philip Papadopoulos 2 1 School of Informatics and Computing, Indiana University

Personal Cloud Controller (PCC)

Yuan Luo1, Shava Smallen 2, Beth Plale1, Philip Papadopoulos2

1School of Informatics and Computing, Indiana University Bloomington2San Diego Supercomputer Center, University of California San Diego

Page 2: Personal Cloud Controller (PCC) Yuan Luo 1, Shava Smallen 2, Beth Plale 1, Philip Papadopoulos 2 1 School of Informatics and Computing, Indiana University

Overview

• Goals:– Enable lab/group to easily

manage application virtual clusters on available resources

– Leverage PRAGMA Cloud tools: pragma_bootstrap, IPOP, ViNE.

– Lightweight, extends HTCondor from U Wisc.

– Provide command-line and Web interfaces

• Working Group: Resources

Page 3: Personal Cloud Controller (PCC) Yuan Luo 1, Shava Smallen 2, Beth Plale 1, Philip Papadopoulos 2 1 School of Informatics and Computing, Indiana University

Client

The PRAGMA Cloud

Cluster A

Cluster B Cluster C

Cluster D

Allocated Resource

Unclaimed Resource

Physical Network

Virtual Network

PCC-HTCondor Master

Provenance Collecting Path

Cluster Master Node

PCC Enabled PRAGMA Cloud

Page 4: Personal Cloud Controller (PCC) Yuan Luo 1, Shava Smallen 2, Beth Plale 1, Philip Papadopoulos 2 1 School of Informatics and Computing, Indiana University

Negotiator

Collector

Startd

Schedd

Central Manager

Startd

Schedd

Shadow

Machine 1 (submit)

Startd

Schedd

Starter

Machine N (execute)

VM GAHP

PRAGMA Cloud tools

(pragma_boot)

Communication PathProcess Invoke

PCC-HTCondor Architecture

Page 5: Personal Cloud Controller (PCC) Yuan Luo 1, Shava Smallen 2, Beth Plale 1, Philip Papadopoulos 2 1 School of Informatics and Computing, Indiana University

PCC-HTCondor Job Submission

universe = vmexecutable = lifemapperlog = simple.condor.logvm_type = rocksrocks_job_dir = /path/to/the/job/dirqueue

executable = pragma_bootbasepath = /opt/pragma_boot/vm-imageskey = ~/.ssh/id_rsa.pubnum_cores = 2vcname = lifemapperlogfile = pragma_boot.log

.vmconf file in the rocks job directory

Sample PCC-HTCondor submission script

Page 6: Personal Cloud Controller (PCC) Yuan Luo 1, Shava Smallen 2, Beth Plale 1, Philip Papadopoulos 2 1 School of Informatics and Computing, Indiana University

Status and Future Plans

• Longer-term goals– Data-aware scheduling– Fault tolerance– Provenance

• Initial prototype implemented– Start and monitor

virtual cluster using pragma_bootstrap via HTCondor (VM GAHP)

– Web interface prototype (PHP)

• Near-term goals – Add increased

controllability and robustness (April – June)

– Multi-site clusters (July – Sept)

Personal Cloud Controller

Rocks

PCC-HTCondor

OpenNebula …

Web Interface

PRAGMA tools (pragma_boot, ViNE, iPOP)

Page 7: Personal Cloud Controller (PCC) Yuan Luo 1, Shava Smallen 2, Beth Plale 1, Philip Papadopoulos 2 1 School of Informatics and Computing, Indiana University

PCC Demo Overview and Setup

1. View PCC Web interfacea. Fully launched

“lifemapper” 8-core virtual cluster

b. Just launched “dock6” 4-core virtual cluster

2. View Condor piecesa. Submit scriptsb. condor_statusc. condor_q

nbcr-224.ucsd.edu• 4 x Dell PowerEdge SC1435

• 2 x Dual-Core 2.4 GHz AMD Opteron

• 8 GB Memory• 250 GB Disk

• Rocks 6.1 with KVM roll• Condor 8.0.6• Pragma_bootstrap + 3 public IPs• PCC + web frontend

vm-container-0-0

vm-container-0-1

vm-container-0-2

nbcr-224.ucsd.edu