grid applications

feelingmomInternet και Εφαρμογές Web

7 Δεκ 2013 (πριν από 3 χρόνια και 6 μήνες)

122 εμφανίσεις

Polish Infrastructure

for Supporting Computational Science

in the European Research Space


Recent

Advancements

of National
Grid

Infrastructure

in

Poland


(
PL
-
Grid
)





A
CC

C
yfronet

AGH,
Krakow
, Poland



Cracow Grid Workshop


CGW’10

11
-
13.10.2010, Krakow, Poland

www.plgrid.pl/en

Jacek Kitowski

2

Acknowledgements


ACC Cyfronet AGH


Kazimierz Wiatr


Marian Bubak


Łukasz Dutka


Karol Krawentek


Andrzej Oziębło


Maciej Malawski


Zofia Mosurska


Robert Pająk


Marcin Radecki


Renata Słota


Mariusz Sterzel


Tomasz Szepieniec


Agnieszka Szymańska


Teresa Ozga




ICM


Piotr Bała


Maciej Filocha


PCSS


Norbert Meyer


Bartek Palak


Krzysztof Kurowski


Tomasz Piontek


Dawid
Szejnfeld



Paweł Wolniewicz


WCSS


Jerzy Janyszek


Bartłomiej Balcerek


TASK


Rafał Tylman


Mścislaw Nakonieczny



3

Outline


General Introduction to PL
-
Grid Project


M
otivation
, consortium and
objectives


PL
-
Grid and EGI


Organization of the project


Current Status and Advancements


Status of hardware infrastructure


Operational Centre


tasks and services


e
-
Infrastructure
s
oftware
, software packages

and
u
ser’s
t
ools


Training and user support


Security aspects


Conclusions


Future work

4

PL
-
Grid Consortium


Consortium

Creation



agreement

signed

January 2007


a
response

to
requirements

of Polish scientists and

due

to

ongoing
G
rid activities
in
Europe
within

the

framework

of
EGI_DS
.


Consortium

members



make
up

of 5
Polish

supercomputing

and
networking

centres
:


Academic Computer Centre CYFRONET AGH, Krakow (coordinator)


Interdisciplinary Centre for Mathematical and Computational Modelling, Warsaw University



Poznan Supercomputing and Networking Centre


Academic Computer Centre, Gdansk


Wroclaw Centre for Networking and Supercomputing


PL
-
Grid

Project
proposal

got

funded

by the European Regional Development Fund

as part of the Innovative Economy
P
rogram


on
March

2, 2009.


Result
:

First
working

NGI
in

Europe
in

the

framework

of
EGI.eu

The Consortium consists of five High Performance Computing Polish Centres

representing Communities
,
coordinated by ACC Cyfronet


Participation in
i
nternational and
n
ational
p
rojects


35+ international projects FP5, FP6, FP7 on Grids (~50% common)

(~45 currently, including EDA)


15+ Polish projects (~50% common)


Polish
s
cientific
c
ommunities


~75% highly rated Polish publications in 5 Communities


Computational resources


Top500 list


European/Worldwide integration Activities



EGEE I
-
III, EGI_DS, EGI, e
-
IRG, PRACE, DEISA, OMII, EU Unit F3 „Research Infrastructure”
Experts, Representatives in EGI Council, Members of EGI Executive Board


National Network Infrastructure
available


Pionier National Project

5

Rationales behind PL
-
Grid Consortium

Rank

Site

System

Cores

R
max
(TFlops)


R
peak
(TFlops)

161

Cyfronet

AGH Krakow


Cluster

Platform 3000 BL2x220, L54xx 2.5
Ghz
,
Infiniband

/ 2010 Hewlett
-
Packard

6144

39.93

55.54

181

Gdansk University of
Technology, CI Task


ACTION
Cluster

Xeon E5345
Infiniband

/
2008, Action

5336

38.17

49.73

444

PCSS Poznan


Cluster

Platform 3000 BL 2x220, E5530 2.4
GHz
,
Infiniband

GDR / 2010, Hewlett
-
Packard

3456

26.22

31.24

TOP500


June

2010

6

PL
-
Grid Project
-

B
asic
D
ata



The Project is co
-
funded by the European Regional Development Fund as part
of the Innovative Economy
P
rogram
.


Total budget: 83 M PLN (~
21 M EUR
)


Personel cost 27 M PLN (~
7 M EUR
)


Equipment cost 33 M PLN (~
8 M EUR
)


Other cost 23 M PLN (~
6 M EUR
)


Funding from the EC: 68 M PLN (~
17 M EUR
)


Project duration:
01 January 2009


31 December 201
1


Beneficiary:
Academic Computer Centre Cyfronet AGH
,
Krakow, Poland


Contract number: POIG.02.03.00
-
00
-
007/08



Project website:


www.plgrid.pl/en

7

Grid infrastructure (Grid services) PL
-
Grid

Application

Application

Application

Application

Clusters

High Performance Computers

Data repositories

National Computer Network PIONIER

Domain

Grid

Advanced

Service

Platforms

Domain

Grid

Domain

Grid

Domain

Grid

Main Objectives of
PL
-
Grid


Polish Grid
is developing
a
common base infrastructure



compatible and interoperable with European
and Worldwide Grids


Capacity to construct s
pecialized,
domain Grid systems


including services and tools focused on
specific types of applications


This
approach should enable
efficient use
of available
financial resources


Plans for
HPC and Scalability Computing
enabled with more focus on domain specific Grids


Offer for the Users


Computing
Power

215 Tflops


Storage

2500 TBytes


Support

from PL
-
Grid staff

on using advanced Grid tools


Support

on porting legacy codes

to Grid environment


Support

on designing applications

for PL
-
Grid environment


8

PL
-
Grid
Building Blocks

PL
-
Grid software

comprise
s
:


user tools

(portals, systems for
applications management and
monitoring, result visualization and
other purposes, compatible with the
lower
-
layer software used in PL
-
Grid)


software libraries


virtual organization systems
:
certificates, accounting, security
,
dynamic


data management systems
: metadata
catalogues, replica management, file
transfer


resource management systems
: job
management, applications, grid
services and infrastructure
monitoring, license management,
local resource management
,
monitoring


Users

National

computer

network

Grid
Application
Programming
Interface

Virtual organizations and

security systems

Basic Grid
services

Grid

services

LCG/gLite

(EGEE)

UNICORE

(DEISA)

Other

Grids

systems

Grid

resources

Distributed

computational
resources

Grid portals, development tools

Distributed
data
repositories

T
hree Grid structures

are maintained:


production


research


development / t
esting


9


EGI.eu



European
organi
z
ation
being developed to coordinate the
European Grid Infrastructure, based
on the federation of individual
National Grid In
itiatives

(NGI), to
support a multi
-
disciplinary user
community.


PL
-
Grid tasks in EGI

(planned/fixed)


Grid operation and oversight of the
e
-
Infrastructure


Coordination of resource allocation
and of brokering support for VOs
from NGIs


Computational Chemistry


organization and management of
Computational Chemistry and
Material Science and Technology
Specialized Support Centre and EGI
liaisons


Development of
u
nified
m
iddleware
via European Middleware Initiative


PL
-
Grid


first operational NGI in Europe

Integration Activity

in the framework of European Grid Infrastructure


S
cientific application porting, especially concerning UNICORE architecture, within Application
Porting SSC

PL
-
Grid

On 30 March 2010, Poland
-

as the first country in
Europe
-

has initiated functioning of the National
Grid Initiative (NGI)

10

Organization of the PL
-
Grid project

PLANNING AND

DEVELOPMENT

OF INFRASTRUCTURE

P2

Coordination

Strategic Planning

Dissemination

PROJECT MANAGEMENT

P1

SECURITY CENTER

P6

Training

SUPPORT FOR VARIOUS

DOMAIN GRIDS

P5

P4

GRID SOFTWARE

AND USERS

TOOLS DEVELOPMENT

EGEE

DEISA

… .

OPERATIONS CENTER

P3

Current

Status and
Advancements

12

13

Status of Hardware
Infrastructure

Computational power


CYFRONET:
29,3 Tflops


ICM: 13,1 Tflops


PCSS: 23,2 Tflops

Disk storage


CYFRONET: 313

TBytes



ICM:

433
TBytes


PCSS:

300
TBytes

Total:
65,6 Tflops

Total:

1046 TB
ytes


Plans until the end of 2010:

Plans until the end of the Project
(end of 2011):

185 Tflops

215 Tflops

1900 TBytes

2500 TBytes

Current status (October 2010):

14

Operational Center


just in operation

Tasks


Coordination of
o
peration
s


Management and accounting


Collaboration with
EGI and
PRACE/
DEISA


Users’ requirements analysis for
operational issues


Running infrastructure for:


Production


Developers


Research


Future consideration:


Computational Cloud


Data Cloud


Internal and External

Clouds


Virtualization aspects

15

Services of the Operational Center for Users


Operational Center aims at facilitating access to the infrastructure by simplifying the
procedures and deployment of useful tools:


System of registration of account management
of the PL
-
Grid user


available at https://konto.plgrid.pl/


required entry in the Polish database of „People of Science” or confirmation
of the scientific tutor


grid access to PL
-
Grid resources


5 centers


gLite


1 center


UNICORE


local access to the queue system


„zeus” cluster in ACC CYFRONET AGH


ability of application for a grid certificate on
-
line


application for access to computational services in other centers


Helpdesk system in PL
-
Grid


enables reporting and tracking issues


available at
https://helpdesk.plgrid.pl

(for registered users)


access also by e
-
mail: helpdesk@plgrid.pl


manual:
https://wiki.plgrid.pl/doku.php?id=pakiet5:publiczne:podrecznik_uzytkownika_pl
-
grid → System
Pomocy Helpdesk

16

Services of the Operational Center for Users (cont’d)


Operational Center cares for a proper functioning of the infrastructure for PL
-
Grid
users by pro
-
active monitoring of the following infrastructure elements:


availability of the infrastructure services


software packages supported by PL
-
Grid


Provision of the conformity of the PL
-
Grid and European (EGI) infrastructures


software


operational procedures


security procedures


Advanced work on the „PL
-
Grid grants” idea


Integration of the data presentation concerning resources usage for a user


Work on provision of the integrated user portal

17

Development of the e
-
Infrastructure Software

and the User’s Tools



Close cooperation of 8 programming and testing groups, about 20 people


Installation and provision for testing purposes gLite, Unicore and QosCosGrid


About 30 various configurations of virtual machines with installed software used for
development and testing of the tools for users


the choice of the technology made


Functional, conformity and efficiency tests of selected packages of the research
software made in order to perform the deployment and support of the new tools and
services on the production level

18

Development of the e
-
Infrastructure Software

and the User’s Tools

(cont’d)


Direct contact with new users on the basis of a survey, available at:
www.plgrid.pl/ankieta



Requirements of the Polish users (results ~190 surveys) considered in the new
applications, tools and services developed and tested in the framework of the
Package 4


Large group of users cooperating with the software and tools team


Department of Chemistry of the Jagiellonian University


Department of Bioinformatics and Telemedicine of the Collegium Medicum of
the Jagiellonian University


Uni
versity of
Adam Mickiewicz


Pozna
n Technical University


Wrocław Technical University


Administrato
rs of the computing centers




19

Development of the e
-
Infrastructure Software

and the User’s Tools

(cont’d)


Test versions of tools for users and systems administrators:
Grid Resource Bazaar
, mobile access to the infrastructure,
new security modules


Extending of the
GridSpace2

platform with a set of new
functions, support for new scripting languages and
integration with new grid services


Integration of the
Migrating Desktop
and

gEclipse

tools
with various PL
-
Grid middleware services


Extension and deployment of
FiVO



a new tool for VO
management and monitoring


Performance and functional tests of middleware service
QosCosGrid

and integration with g
Lite
and

Unicore

infrastructure at the queue systems level


Implementation and provision of the advanced graphical
interfaces, visualization and tasks and data management for
selected user applications via the
Vine Toolkit


Integration of the selected tools and web applications with
Liferay portal framework and Nagios monitoring system

20

Grid Resource Bazaar


Providing resources to users with required
qualities of services


Required = specified in Service Level Agreement


RU



R
esource
U
sers

RP



R
esource
P
roviders

21

Grid Resource Bazaar (cont’d)

Resource Allocation
-
related Operation Model

22

http://grid.cyfronet.pl/bazaar

Grid Resource Bazaar (cont’d)

23

GridSpace2 Experiment Workbench in the PL
-
Grid
Virtual Laboratory



Use of distributed computational resources and data repositories


High
-
level tools offered for the user for in
-
silico experiments

24

GridSpace2
-

w
orking with the Experiment
Workbench

The PL
-
Grid Virtual Laboratory provides
an
Experiment Workbench


a Web 2.0
interface supporting the development and
execution of in silico experiments.

W
ork
ing

with the Experiment Workbench:


Open the workbench
in your browser
(
https://wl.plgrid.pl
)


Log into

one of the available servers with your PL
-
Grid account


Start your usual work
. Your files are already there.
Your code snippets may play the role of scripts (in
Python, Perl, Ruby etc.), bash commands or input to
external applications (such as Gaussian or
Gnuplot
);


All the
files

you generate
may be viewed with
visualization tools

(such as
Jmol
)


Save your work



the experiment is now visible
among your other files


Share the experiment
with other members of your
research team

25

GridSpace2
-

sample experiment in
ViroLab

environment


Patient’s data


Medical examination


HIV genetic sequence put into
database


Experiment
in
-
silico


Collect HIV genetic sequences
from database


Perform sequence matching



Calculate virus resistance


http://
gs2.cyfronet.pl/

http://www.virolab.org


26

Migrating Desktop Platform


the Framework for Grid
Applications

M
ain

product features
:


Simple
integration

with
grid applications


Easy job defining, submission, monitoring and visualization of results


Support for
batch

and
interactive jobs


Handling of
sequential

and
parallel applications


Intuitive
management of grid data


Easy
extendable framework

Migrating Desktop main window

Description
:


The Migrating Desktop Platform is a
powerful
and flexible user interface

to Grid resources
that gives a transparent user work environment
and easy access to resources and network file
systems independently of the system version
and hardware


It allows the user to run applications and tools,
manage data files, and store personal settings
independently of the location
or the terminal

27

Migrating Desktop Platform


the Framework for Grid
Applications

(cont’d)

Application support
:


The
key feature
:

the possibility of
easy
adding various tools, applications

and
supporting

different
visualization formats


The Migrating Desktop offers
a framework
that can be easily extended
on the basis of a
set of well
-
defined plug
-
ins used for: accessing
data, defining job parameters, pre
-
processing
job parameters, and visualization of job results


Open architecture
of the Migrating Desktop
speeds up the application integration process
and makes that product significantly
more
flexible than specialized tools

(e.g. portals)
designed only for a specific application

Job defining and
visuali
z
ation of
results (examples
of Migrating
Desktop plug
-
ins)

28

g
-
Eclipse


to Access the Power of the Grid

Benefits from using g
-
Eclipse:


Grid application users will benefit from the
desktop
-
like access
to Grid resources


Grid operators and resource providers will be able
to
reduce the time
-
to
-
service

by using the Grid
management and Grid site configuration tools


Grid application developers will
reduce the time
-
to
-
market
for new Grid applications by accelerating
the development and deployment cycle

Features
:


The g
-
Eclipse framework provides tools to
customi
z
e Grid users’ applications, to manage
remote resources and to support the development
cycle of new Grid applications


The framework
consists

of
general Grid
workbench tools that can be extended for many
different Grid and Cloud middleware’s (such as
gLite
, UNICORE,
Globus

Toolkit, Amazon Cloud)

29

g
-
Eclipse


to Access the Power of the Grid

(cont’d)

g
-
Eclipse elements (functionalities)
:


Grid Virtual Organi
s
ation management
allows
the dynamic creation and management of Virtual
Organisations and their resources


Grid job management
supports Grid users and
developers
in

creat
ing

and manag
ing

Grid jobs
independent of the middleware


Grid file management
provides access and
management of local and foreign resources
which
are seamlessly integrated into the Eclipse
Workbench


Grid application deployment
supports Grid
application developers and Grid application users
with the deployment of their applications on the Grid

g
-
Eclipse provide
s

API and User Interface
modules for all
above

functionalities.


Grid
visualisation

tools
offer functionalities that allow the
visualisation

of
scientific

and numerical
calculations


Grid
workflow

builder
support
s

the creation and management of middleware independent
workflows

by
use

of a

provided

graphical editor


Grid command console
is a command line interface to access the Grid

30

FiVO/QStorMan toolkit


the goals


Provisioning PL
-
Grid users an appropriate quality of access to storage

systems with the following means:


Definition of the non
-
fu
n
ctional requirements for the storage systems, e.g.
expected transfer rate of the read/write operations or parameters related to
expected availability level of the resource or security.


Monitoring of the storage systems parameters, e.g. a distributed file system.


Data localization management based on the provided requirements and the
current state of an environment.


QStorMan is a part of the Framework for Inteligent Virtual
Organization (FiVO)

31

FiVO/QStorMan toolkit
-

architecture

and components

Components of the toolkit:


Portal



a graphical user interface for
defining requirements for storage systems



GOM

(Grid Organizational Memory)




a knowledge base for storing semantic
description of the storage systems along
with defined requirements



SMED



a knowledge
-
supported

monitoring

system



SES
(Storage Element Selection)
library


a programming library for selecting
storage resource based on provided
requirements and current state of the
environment


SE


Storage Element

32

QCG


QosCosGrid Middleware

Description:


The QosCosGrid (QCG) middleware is
an integrated e
-
infrastructure
offering advanced
job and resource management
capabilities to deliver to
end
-
users
supercomputer
-
like performance

and structure


By connecting many computing clusters together, QosCosGrid offers easy
-
to
-
use
mapping, execution and monitoring capabilities
for variety of
applications, such as parameter sweep, workflows, MPI or hybrid MPI
-
OpenMP

QCG
architecture

33

QCG


QosCosGrid Middleware

(cont’d)

Benefits:


for
End Users



efficient and secure access to
dynamically discovered computational resources
located in the PL
-
Grid environment


for
Cluster and Grid Administrators



a great
opportunity to share and use their clusters more
efficiently by scheduling local and Grid end users


applications according to defined policies and
constraints


Several
portal based
and
mobile applications
have
been developed and integrated with QCG
infrastructure to give the users intuitive
-

experiment
and domain oriented
-

access to resources.

QCG mobile client

34

Vine Toolkit


Distributed Resources at Your
Fingertips for Developers and End Users

Main features:


Single Sign
-
On


users log in once and use all resources


Support for different grid infrastructures


Advanced web applications (integration with Adobe Flex / BazeDS technology)


Extensible, module based open architecture based on plugins


Many applications included „out
-
of
-
the
-
box” such as: Job Manager, File Manager,

Resource Manager, Credential Manager, Certificate Manager, GSI
-
SSHTerm applet


Support for scientific applications: ABINIT, Quantum Espresso (and others…)


I
ntegration with portal environments Liferay and GridSphere installed on Tomcat server

General description:


Vine Toolkit was designed as an environment to
facilitate the development
and integration of web
-
based applications
with HPC resources, Grid
services and various existing large
-
scale computing infrastructures managed
by Grid middleware, such as gLite, Unicore, Globus, QosCosGrid and GRIA.


It is possible to
create a computational web
-
based Science Gateway

using
a modular structure and existing Vine components.

35

Vine Toolkit


Distributed Resources at Your
Fingertips for Developers and End Users

(cont’d)

For whom:


web applications developers




a base for
building
advanced
computational scientific portal
gateways and

applications
with
many ready to use components,
integration layer for portal
frameworks and advanced

installation engine


administrators of grid
infrastructures


possibility of
deployment of a web portal
“out
-
of
-
the
-
box”

to access

existing grid infrastructure


end users


many
ready to use
web applications

for grid
computations and data
management in grid

infrastructure


scientists



dedicated
web
applications
for scientific
applications for different scientific
domains

Example of the advanced web interface in the
Nano
-
Science
Gateway based on Vine Toolkit

36

Scientific software packages


Access to software packages is provided to users through:


gLite


UNICORE


Examples of available packages in various fields:


biology
: AutoDock, BLAST, ClustalW2, CPMD, Gromacs, NAMD


quantum chemistry
: ADF, CFOUR, Dalton, GAMESS, Gaussian, Molcas, Molpro,
MOPAC, NWChem, OpenBabel, Siesta, TURBOMOLE


physics
: ANSYS FLUENT, Meep


numerical computations and simulation
: Mathematica, MATLAB


other
: Blender, POV
-
Ray


Users may report us their expectations through a survey available at:

http://www.plgrid.pl/ankieta


The system of testing the software packages in the grid environment has been
prepared and deployed


The correctness of functioning of the packages is monitored automatically in each
of the centers

37

Training and Users’ Support


Basic training

on access to the PL
-
Grid infrastructure through gLite and
UNICORE conducted in all centers participating in the project


in Gdańsk /
Kraków / Poznań / Warszawa / Wrocław


More advanced training started


Similar (free) training may be conducted in other centers, if necessary


eLearning courses

offered through the Blackboard system ( available for
registered users of the PL
-
Grid infrastructure)


Helpdesk system

implemented


it’s a novel support system for people using the Project resources


it involves the technical support and organization of the current users’
support by the experts (maintenance of trouble tickets)


tickets in Helpdesk may be created by sending an email to:
helpdesk@plgrid.pl

or through the online system, available at:
https://helpdesk.plgrid.pl



38

Security in
PL
-
Grid



Provision of two CAs


PKI certification centers


for grid users


Project and implementation of the SimpleCA system, facilitating the
users obtaining PKI certificates and their usage


Project and implementation of the secure configuration of the
infrastructure, in conformity with the most actual security standards


Project of the system monitoring the conformance of the
configuration deployed in the centers with the security policy


Creation of the group of experts from the field of security, in order to
continuously monitor the environment, immediate react on incidents
and support users and administrators


Prototype version of the
system of correlation of
information about
the attacks on the infrastructure (ACARM
-
ng)


Audits of applications crucial for grid security

Conclusions


Good and promising partial results of the PL
-
Grid Project


Justified expectations to Project completion according to its goals and
requirements


Generic services and generic applications in development



Futher developement needed, as identified currently, mainly on Domain
Specific Grids


Request from the users’ communities


Capacity for organization of future development according to


Expertise and experience


Strong scientific potential of the users’ communities being represented
by PL
-
Grid Partners


Wide international cooperation concerning the Consortium and
individual Partners, good recognition worldwide


Good managerial capacity

39

40

http://www.plgrid.pl/en