John Kewleye-Science Centre

All Hands Meeting 200521st September, Nottingham

GROWL: A Lightweight Grid Services Toolkit and Applications

John KewleyCCLRC Daresbury Laboratory

j.kewley@dl.ac.uk

Presenter NameFacility Name

John Kewleye-Science Centre

All Hands Meeting 200521st September, Nottingham

Grid Middleware

"We have encountered serious middleware-related problems which are hindering scientific progress with the Grid:

The existing toolkits have an excessively heavy set of software and administrative requirements, … ;

Existing toolkits are painful and difficult to install …;

Existing standards bodies … are not engaging sufficiently with the applications community, and run a substantial risk of producing and implementing Grid architectures which are irrelevant to the requirements of application scientists."

Chin and Coveney, RealityGrid, 2004

Presenter NameFacility Name

John Kewleye-Science Centre

All Hands Meeting 200521st September, Nottingham

Installation Difficulties

Typically need to be root to install

Software must be downloaded from various locations (unless software stacks such as OMII / VDT is used)

There are many choices for type of installation (too many options?)

Firewalls can get in the way when setting things up

Presenter NameFacility Name

John Kewleye-Science Centre

All Hands Meeting 200521st September, Nottingham

GROWL

Collaborative project (JISC VRE programme) between CCLRC Daresbury Laboratory and the Universities of Cambridge and Lancaster.

Project Objectives: to produce a lightweight client-side Grid connection toolkit.

Presenter NameFacility Name

John Kewleye-Science Centre

All Hands Meeting 200521st September, Nottingham

GROWL Application AreasBioinformaticsAnalysis of microarray expression data. (R and Matlab are popular)

Computational ChemistryCollaboration with CCP1 project to incorporate GROWL C library calls with legacy code, e.g. DLV, GAMESS-UK ,

Social ScienceGrid-enable the SABRE statistical modelling package (based on R) used for work/life history data analysis.

http://www.growl.org.uk/

Presenter NameFacility Name

John Kewleye-Science Centre

All Hands Meeting 200521st September, Nottingham

GROWL will provide Transparent client-side handling of Grid-related

issues e.g security, file transfer etc.

Modules, libraries and wrappers that interface with existing client software tools

Extensibility via a simple API with common language mappings (C++, C and Fortran).

A persistent multi-client server linked to existing grid components (primarily the Globus toolkit) providing access to HPC resources, session management, scheduling, authentication etc.

Presenter NameFacility Name

John Kewleye-Science Centre

All Hands Meeting 200521st September, Nottingham

But what does that mean?

Basically, GROWL will only download and build sufficient software to provide a client-side interface to Grid software.

It will not build you a grid resource!

GROWL will provide simple, generic interfaces to Grid middleware - remember the 80/20 rule!

If you can have C-linkable libraries in your favourite GIS package, you should be able to link to GROWL (e.g. Matlab, Stata, R).

Presenter NameFacility Name

John Kewleye-Science Centre

All Hands Meeting 200521st September, Nottingham

How GROWL works

Presenter NameFacility Name

John Kewleye-Science Centre

All Hands Meeting 200521st September, Nottingham

GROWL Installation

Only builds the GROWL modules you request

Downloads and installs any dependent software for you: e.g. Globus, SRB client, gSOAP, MyProxy

Currently Linux only, but aiming to support Windows XP and other Unix

Presenter NameFacility Name

John Kewleye-Science Centre

All Hands Meeting 200521st September, Nottingham

The Modules +Wrappers

Modules Authentication / Session Management Job submission: currently Globus, but working on Condor. File movement: including 3rd party SRB client

Wrappers R (including from Sabre) Intend to add Matlab and Stata

Presenter NameFacility Name

John Kewleye-Science Centre

All Hands Meeting 200521st September, Nottingham

Certificate handling

1. User's e-Science Certificate2. Local proxy certificate3. Certificate on MyProxy server for delegation to

Growl Server

Second two can be done in one go without endless asking for and asking for confirmation of passwords.

Presenter NameFacility Name

John Kewleye-Science Centre

All Hands Meeting 200521st September, Nottingham

Certificate tool

$ openssl pkcs12 -in mykey.p12 \ -clcerts –nokeys \ -out usercert.pem <Pass1> <Pass2> <Pass2> [confirm]

$ openssl pkcs12 -in mykey.p12 \ –nocerts -out userkey.pem <Pass1>

$ chmod 444 usercert.pem

$ chmod 400 userkey.pem

$ … .globus/ …

$ mk_cert mykey.p12 <Pass1> [<Pass2>]

Presenter NameFacility Name

John Kewleye-Science Centre

All Hands Meeting 200521st September, Nottingham

Proxy production tools

$ myproxy-init <Pass2> <Pass3> <Pass3> [confirm]

$ grid-proxy-init <Pass2>

$ # Needs extra setup

$ grid-login <Pass2> <Pass3>

Presenter NameFacility Name

John Kewleye-Science Centre

All Hands Meeting 200521st September, Nottingham

Further Information

1. Sabre-R and Growl Session Management (Dan Grose, University of Lancaster, Video)Uni of Cambridge e-Science Centre

2. MCMC Simulation (Lorna Morris)Uni of Cambridge e-Science Centre

3. Certificate scripts and/or Growl BuildOn request

Presenter NameFacility Name

John Kewleye-Science Centre

All Hands Meeting 200521st September, Nottingham

In Summary

1. Usability (Simple things should be easy!)

2. Accessibility (R, Stata, Matlab, Fortran, …)

3. Lightweight

4. Strong links with User communities