AUSTRIAN GRID - RISC - JKU

flameluxuriantΔιαχείριση Δεδομένων

16 Δεκ 2012 (πριν από 4 χρόνια και 10 μήνες)

392 εμφανίσεις

A
USTRIAN
G
RID





1
/
16

A
USTRIAN

G
RID


Report on the Use of Globus 4 and the Web Service
Resource Framework for SEE++


Document Identifier:

AG
-
DA
-
1c
-
4
-
2006_v1.doc

Status:

Public

Workpackage:

A1c

Partner(s):

Research Institute for Symbolic Computation (RISC)

Upper Austrian Rese
arch (UAR)

Lead Partner:

RISC

WP
Leader
s
:

Wolfgang Schreiner (RISC), Michael Buchberger (UAR)


A
USTRIAN
G
RID





2
/
16


Delivery Slip


Name

Partner

Date

Signature

From

Karoly Bosa

RISC

2006.
11
.31



Verified by






Approved by







Document Log

Version

Date

Summary o
f changes

Author

1.0

2006
-
11
-
3
1

Initial Version

See cover on page 3














A
USTRIAN
G
RID





3
/
16






Report on the Use of Globus
4 and the
Web Service Resource Framework

for
SEE++





Karoly Bosa

Wolfgang Schreiner


Research Institute for Symbolic Computation (RISC
)

Johannes Kepler University Linz

{Karoly.Bosa, Wolfgang.Schreiner}@risc.uni
-
linz.ac.at



Michael Buchberger

Thomas Kaltofen


Department for Medical Informatics

Upper Austrian Research (UAR)

Thomas.Kaltofen@uar.at



März 18, 2013



A
USTRIAN
G
RID





4
/
16


DELIVERY SLIP

................................
................................
................................
................................
...................

2

DOCUMENT LOG

................................
................................
................................
................................
................

2

1

ABSTRACT

................................
................................
................................
................................
.................

5

2

INTRODUCTION

................................
................................
................................
................................
.......

6

3

A WS
-
GRAM COMPATIBLE EXTE
NSION OF SEE++

................................
................................
.......

7

3.1

T
HE NEW
J
OB
D
ESC
RIPTION

................................
................................
................................
..................

8

3.2

U
SER
I
NTERFACE

................................
................................
................................
................................
..

9

3.3

B
ENCHMARKS

................................
................................
................................
................................
.....

10

4

USING THE PBS

SCHEDULER FOR SEE++
JOB SUBMISSIONS

................................
..................

11

5

EXPERIENCES WITH SEE
++ DATA MANAGEMENT S
YSTEM BASED ON WSRF

.................

11

5.1

D
ISCOVERED
L
IMITATI
ONS

................................
................................
................................
.................

12

5.2

A

P
RELIMINARY
E
XPERIMENTS WITH THE
WSRF

B
ASED
M
EDICAL
D
ATABASE
S
ERVICES

.................

12

6

OUTLOOK

................................
................................
................................
................................
................

13

7

ACKNOWLEDGEMENTS

................................
................................
................................
......................

15

REFERENCES

................................
................................
................................
................................
....................

15


A
USTRIAN
G
RID





5
/
16

1

Abstract


SEE
-
GRID is based on the SEE++ software system for the biomechanical simulati
on of the
human eye. The goal of SEE
-
GRID is to extend SEE++ in several steps in order to develop an
efficient grid
-
based tool for “Evidence Based Medicine”
, which supports surgeons in
choosing optimal surgery techniques for the treatment of certain eye mo
tility disorders.


First, we have developed a grid
-
enabled version of the simulation of the Hess
-
Lancaster test,
which is a medical examination by which the pathology of the patient can be estimated. Based
on this, we work on a pathology fitting algorithm
that attempts to give sufficiently close
estimations for the pathological reasons of the disorder. Furthermore, we have developed a
prototype version of a grid enabled medical database where both real and simulated
pathological cases can be collected, sort
ed and evaluated for improving both the later
pathology fitting calculations and the future medical treatments.


In this document, we present an extension of the “SEE++ to Grid Bridge”, with which
it is
able to start

SEE++ server

processe
s

on a

grid
site
v
ia WS
-
GRAM. Then we
describe

how
“Grid
-
Enabled SEE++” can work together
with
the OpenPBS batch scheduler, w
hich is
deployed on some
cluster
s

of the Austrian Grid.



Finally,
w
e
give an analysis about our experiences with

integration of the prototype Web
Se
rvice ba
sed imp
lementation

of the SEE
-
GRID database into the WSRF framework
.

A
USTRIAN
G
RID





6
/
16

2

Introduction




Figure 1. The Extended Architecture of the „Grid
-
Enabled SEE++“

based on Globus 4


“Grid
-
Enabled SEE++”

is ba
sed on the SEE++ [SEE
-
KID, 2006;
Buchberger, 2004
;
K
altofen, 2002
]

software system

for the biomechanical 3D simulation of the human eye and

its muscles.

SEE++ simulates the common eye muscle surgery techniques in a graphic

interactive way that

is familiar to an experienced surgeon.

The goal of

“Grid
-
Enabled

SEE++”

is to adapt and to extend SEE++ in several steps and

to develop an efficient grid
-
based

tool
for “
Evidence Based Medicine

, which supports the surgeons in

choosing optimal surgery
techniques

for the treatments of different syndromes of strabismus.


In [
Bosa
, 2006]
, we combined the SEE++ software with the Globus (pre
-
Web S
ervice)
middleware [Globus, 2006]

and developed a parallel vers
ion of the simulation of the
Hess
-
Lancaster

test

(typical medical examination). By this, we demonstrated how a noticea
ble
speedup can be achieved in SEE++ by the exploitation of the computational power of the
Austrian Grid. Furthermore, we reported the prototype implementation of a medical

database
component for “Grid
-
Enabled SEE++”
.

Finally, we design
ed a so called grid
-
based
Pathology
Fitting

algorithm,
which would be able to determinate (or at least estimate) automatically

the
pathological reason of a patient's strabismus.


The
current architecture of “Grid
-
Enabled SEE++”

(the box in Figure 1

bordered by the
dashed line
) consists of the Web Service based database services, WSRF
-
based
database
services, the “SEE++ to Grid Bridge”
, the grid
-
enabled SEE++ servers (which are started via
A
USTRIAN
G
RID





7
/
16

pre
-
WS GRAM and perform the gaze pattern calculations) and the SEE++ clients. Accor
ding
t
o the

scenario

of the parallel simulation of

Hess
-
Lancas
ter, b
efore the bridge accepts the

computational requests from the SEE++ clients, it submits in advance
some grid
-
enabled
SEE++ servers into the grid. These processes behave as some kind

of “
executer


processe
s
for the computation tasks in order that the remarkable latencies of the job submissions in case
of the computational requests can be avoided

(since the
developed
parallel Hess
-
Lancaster
test simulation takes only

approximately 1 up to 45

secon
ds on the grid).


T
he SEE++
clients can connect to all database

component
s located on the grid via the
“SEE++ to Grid Bridge”

(the underlying grid
-
based infrastructure is hidden from the clients).
Furthermore, the clients can also reach every Web Service b
ased database component via the
bridge, although a client is also
able to interact only with one such database directly.

For the
underlying database system, we decided to use My
SQL, because in our comparative
performance tests it worked together with our d
eveloped software architecture 4
-
8 times faster
than

postgreSQL.


Recently,
w
e
have just finished
an extension of the “SEE++ to Grid Bridge”, with which it is
able to start grid
-
enabled SEE++ servers on a grid site via WS
-
GRAM, see Section 3.
Furthermore,
we describe in this document, how “Grid
-
Enabled SEE++” can work together
with the OpenPBS batch scheduler, which is deployed on some clusters of the Austrian Grid,
see Section 4.


Finally, w
e attempted to ref
ine

the elementary integration of the prototype
Web Service ba
sed
implementation [Mitterdorfer, 2005]

of the SEE
-
GRID database into the WSRF framework
.

But since we met very strict limitation
s

from
the
side of C WS Core of Globus 4, we could
make only preliminary experiences with WSRF
-
based medical data
base of SEE++,
see
Section 5
.


3

A WS
-
GRAM Compatible E
xtension of
SEE++


Until now, we could not implement a fully functioning WS
-
GRAM client using the
C WS
Core API
and the
C WS
-
GRAM API
, a
lthough

we

spent lots of time with reading of their
available docu
mentations

and discussed the situation with some other resea
rch groups of the
Austrian Grid.

We found there is simply not e
nough documentation for these APIs
. The only
complete

and
well
function
ing ANSI C
client
application for WS
-
GRAM,
whose source is
ava
ilable
,

is the G
lobus tool
globusrun
-
ws

which
is a very general and
complex software. So
we decided

to

call this
applic
ation from our “SEE++ to Grid B
ridge


in order to submit our
SEE++ server processes via WS
-
GRAM.
By this, we can make some exper
iments

wi
th the
WSRF interface used for this grid

service and we can compare
this solution with the already
used pre
-
WS architecture.
So the globusrun
-
ws is called in the followi
ng form from our
bridge application
:



glo
busrun
-
ws
-
submit
-
S
-
f
<job_description>


-
F https://
<address_of_the_grid_site>
:8443/wsrf/services

/ManagedJobFactoryService


The chosen options are the followings:

A
USTRIAN
G
RID





8
/
16




The


-
submit”
argument submits

or resubmits the job.




The

-
S”
or

-
staging
-
delegate”
will delegate the client credential to WS
-
GRAM
and
Reliable File Transfer
(RFT) service. For SEE++, the
se

are required, because we
apply file stage out (for send back the output file to the client) and clean up after the
job has been executed successfully.




The

-
f”
or

-
job
-
description
-
file”
causes th
e job description to be read from the
given XML file, see Section 3.1.




With

-
F”
or

-
factory”
contact address of the WS
-
GRAM can be given. If this
option is omitted then the job is submitted on the WS
-
GRAM service located on
“localhost“.

The
ManagedJobFa
ctoryService
is one of the available services of WS
-
GRAM which always creates a
n

independent
resource

environment for each
submitted job [Globus, 2006].


Accordingly, t
he "SEE++ to Grid Bridge" acts as a
WS
-
GRAM client and submits a job
described by a
n XM
L based description language, see Section 3.1.



When a SEE++ server is started, it attempts to bind itself to a free port in a predefined port
range and to send

this information back to the “
SEE++ to Grid Bridge

.

Unfortunately this is
not a trivial task,

because
real
-
time
interaction

is not possible

with

a
Job Resource
managed
by WS
-
GRAM.

According to the applied concept, the output files can be stage out from the
grid site if and only if the execution of the corresponding job finished successfully.




Fo
r solving this problem, we have applied a technique that was already

used in [glogin, 2004]
in a similar situation
, but in case of (pre
-
WS) GRAM
. According to this, after the server
process has bo
und itself to a free port, it “forks”

itself and

then termin
ates. The WS
-
GRAM

perceives the termination of the pr
ogram and checks whether there is any predefined
fileStageOut procedure
.

W
e can
use this procedure to
send back the output files to the
“SEE++ to Grid Bridge”.


Fort
unately, once a program has been start
ed on a grid site,
this program will be able to accept
communica
tion requests. Therefore, the “SEE++ to Grid Bridge”

will be able to connect to
the socket, which is still held active by the previously forked copy of the SEE++ server.


3.1

The new Job Descripti
on


Orig
i
nal
l
y,
the Resource Specification Language (RSL) developed for the Globus Project was
used for defining job submission in case of (pre
-
WS) GRAM.
For WS
-
GRAM the job
submissions have to be spe
cified

in a
new
XML based description

language.



<?xml
version="1.0" encoding="UTF
-
8"?>

<job>


<executable>
/location_of_the_executable

/grid
-
seeppserver</executable>

A
USTRIAN
G
RID





9
/
16



<stdout
>/tmp/
seegrid_port_numbers
.txt
</stdout>


<count>
number_of_processes
</count>


<fileStageOut>


<transfer>


<
s
ourceUrl>
file:///tmp/
seegrid_port_numbers
.txt
</sourceUrl>


<destinationUrl>gsiftp://
laddress_of_client
:2811
/
location_of_the_named_pipe
</destinationUrl>


</transfer>


</fileStageOut>


<fileCleanUp>


<deletion>


<fi
le>

file:///tmp/
seegrid_port_numbers
.txt

</file>


</deletion>


</fileCleanUp>

</job>


This is a single job description

template for the submission of the SEE++ server processes
,

which is defined between the root description elements
<job>

and
</j
ob>
. The applied
intermediate description elements are:




executable:
It specifies the location of the executable on the grid site.




stdout:
It defines the standard
output

file for the submitted jobs
.




count:
It

defines the number of processes to execute.




fileStageOut:

In order to send back the output file after job execution, we should add
specific
file staging

e
lements to the job description and we must delegate our
credentials, see Section 3.




fileCleanUp:
Finally,

the
client requires

the clean up proced
ure, in which it denotes
the established temporary or output files for deleting.

For this, the
client credential
must be deleg
ated as well.


3.2

User Interface


At present, i
f the user intends to starts some

SEE++ server process
es

via WS
-
GRAM on a
grid site in
stead of (pre
-
WS) GRAM, then she must start the “SEE++ to Grid Bridge” with the
option “
-
ws”. For instance, i
n case of the following command



A
USTRIAN
G
RID





10
/
16

./seepp2
grid

ws

n 3

altix1.jku.austriangrid.at


where the bridge will start 3 server processes on grid site alt
ix1.jku.austriangrid.at via WS
-
GRAM
.


3.3

Benchmarks


We have compared the overheads of the submission
s

of our SEE++ Server process
es

via pre
-
WS and WS GRAM
in different situations where 1, 3, 9, 25, 30 or 45 pr
ocesses were
executed

on the grid
.


E
ach value in

Table 1

depicts the average

job submission

time of 5
-
7 computations
.

The test
cases were executed on the Austrian Grid
site altix1.jku.austriangrid.at
, which contains 64
Intel Itanium processors (1.4GHz) and resides at the Johannes Kepler University (JKU)

i
n
Linz. The “SEE++ to Grid Bridge”

and SEE++ clients were always
executed at the
RISC
Institute located in Hagenberg which has a one Gigabit/sec connection to

the JKU Linz.


For measuring, we installed the Ethereal network protocol analyzer
[Ethereal, 2
004]

on the
machine where the SEE++ client is executed. By this software, the network traffic between
the local machine and the
mentioned grid site

was filtered and each network package sent to
or received from the port of “seepp2grid” was captured. After
the execution of a test case, the
duration of the job submission

can be determined from the recorded capture time of the first
sent and of the last received message.





Table 1: Comparative Benchmark of the Starting of SEE++ S
erver Processes via (pre
-
WS)
GRAM and WS
-
GRAM


From the measured values contained in Table 1, we can see obvious differences
among

the
overheads of the job submissions in pre
-
Web Service and Web
Service Architectures of
Globus,

i.e. t
he latter is much slow
er
.

On reason for this is certainly

the more robust
architecture and the new
and more sophisticated services. Another reason may

be that t
he
WSRF
-
based s
ervices were implemented in Java

(on the top of the Java WS Core) and not in
C. We hope
that
in the fu
ture versions of Globus the C WS Core will be also extended
to
Machine Name
altix1.jku.austriangrid.at
Number of Submitted
grid
-
Enabled
SEE++ Server
Submission via
(pre
-
WS) GRAM
Submission via
WS
-
GRAM
1
3
9
25
30
45
9,5s
10s
11s
15s
16s
20s
0,85s
0,92s
0,98s
1,06s
1,09s
1,15s
A
USTRIAN
G
RID





11
/
16

those WS

features

which have already been in the Java part, because this could cause a
n

appreciable

improvement in the performance of the Globus Toolkit.


4

Using
the
PBS Scheduler for SEE++ Job
Submissions


Open

Portable Batch System
(OpenPBS) is a computer software job scheduler that allocates
(local) network resources to batch job. OpenPBS is part t
he software specification of Austrian
Grid
[AGRID S. Spec., 2005]

and it is deployed some cluster
s of the Austrian Grid.


Unfortunately, it is quite easy to confuse this local resource management application, if a job
spawns/forks a child process

and terminates. In this case, the scheduler considers that the job
is finished, while the child process i
s still active.
This may induce some problems, because the
batch queueing system assumes the resource is free and may assign some other jobs to it or it
may kill the forked process (in order to clean up). But this is exactly the technique, how
“Grid
-
Enable
d SEE++” is able to send the bound port numbers of it server processes to the
“SEE++ to Grid Bridge”.


Hence, “Grid
-
Enabled SEE++” could not exploit the advantages of the OpenPBS batch
scheduler (on some others e.g.: Torque
-
1

Resource Manager
) for a long t
ime. After some
search we found the paper
[Rosmanith, 2006], which deals with exac
t
ly such weaknesses of
the scheduling system
s

and

it recommend
s

a very simple trick to overcome

the problem
:
we
issue a system call
setsid()
,

which creates a session id for t
he forked process. This helps,
because the OpenPBS tracks the jobs by their session ids.


So if the user would like

to submit some

SEE++ server processes
on cluster
via PBS local
scheduler
, she has to type

the name of the scheduler afte
r the grid site cont
act address, when
she starts the “SEE++ to Grid Bridge”
(if there is not any designated

scheduler or job
manager,

G
lobus will
always
choose the default one, which is the Globus fork
-
jobmanager).


Example:


seepp2grid


n
9

altix1.jku.austriangrid.at
/
jobman
ager
-
pbs


Of co
urse t
here are some other resource management systems (e.g.: Torque
-
2, Condor, etc),
which are not part of the Austr
ian Grid software specification and in which

we cannot avoid
the mentioned problem as easily as
in
this case.



5

Experiences w
ith SEE++ Data Management System
Based on WSRF



In
[SEE
-
GRID, 200
6/3], we have already presented the initial version of the WSRF based
medical database for SEE++.

Now we
have
started to extend the “SEE++ to Grid Bridge”
with the client side functionality

for this database service.


A
USTRIAN
G
RID





12
/
16

5.1

Discovered Limitations


Since the “SEE++ t
o Grid Bridge” is implemented in C/C++, we should apply the APIs of the
“Globus C WS Core” for implementing the client side functionality our database service.


First, we should genera
te client side stub files from the WSDL file. For this purpose, we
applied the
globus
-
wsrf
-
cgen

tool
. Unfortunately, this tool has some strict limitations, too:




It o
nly generates bindings from
Document/literal
style WSDL schemas
.

Since our
original WSDL f
ile used
RPC/encoded

style, we had to regenerate the file with
Document/literal (we did it before we started to create the java stubs for the server side
of course).




It o
nly generates ANSI
-
C bindings. C++ bindings are not supported.

The solution
for this
problem is not as easy as in the previous case. We employed a very complex
data structure as arguments in the SOAP messages, whose bindings were im
plemented
in C++ on the “SEE++ t
o Grid Bridge” (which implementation was taken from the
original SEE++ softwa
re). Therefore, we must implement a conversion between the
generated ANSI
-
C bindings and the already applied C++ data structure (since we do
not want to re
-
implement the whole “SEE++ to Grid Bridge” based on only the ANSI
-
C data structure).


We were

most
surprised
, when we checked the generated ANSI C stubs. M
ost of the data
structure

of our

eye model structure was omitted
from the generated C file. After

a little
while, we realized that impleme
ntation of the data structure

called

SOAP Encoded Array


is
c
ompletely missing from the C WS Core of Globus
. Since our applied eye model

strongly
depend
s on this

data type, we cannot apply
the
WSRF framework in our software system
. I
t
also
does not make sense to re
-
implement the whole eye model without SOAP arrays,
because
the
development of “Grid
-
Enabled SEE++” would diverge from the development line of the
original SEE++, what could leads se
rous incompatibility problems between

future versions of
the
se

software packages
).


5.2

A
Preliminary E
xperiments with the

WSRF Ba
sed Medical
Database S
ervices


We investigated as an alternative solution to use gSOAP [gSOAP, 2005] to interact with the
WSRF
database component (which do
es

not use any special featu
res of Globus 4 at the
moment). We chose this way
, because we rely on tha
t

we are able to

make some preliminary
performance tests with our WSRF database serv
ice with only a little effort (s
ince SOAP is a
platform
-

and language
-
independent
protocol
and we
can
use
on the client side
gSOAP as
before, we do not need to re
-
implement

our C++ data structures and functions).



Of course, these are

very artificial test circumstances, since obviously we cannot start and use
the

Globus Web
Service Container
” in usual way
. This means we must start the container
with the “
-
nosec” option:


g
lobus
-
start
-
container
-
nosec

A
USTRIAN
G
RID





13
/
16


This
lets the container accept and send

only
plain SOAP messages without
any SSL
encryption. Furthermore, grid
-
style certificates are not applied as well (so there is no security
at all). Hence, we cannot perform the test on r
eal grid site of the Austrian Grid

of course
, but
only in our local test environments.


For the test
,

we regenerated the client
-
side

stubs

by gSOAP
from WSDL file used
already in
our WSRF based database and
recompiled the “SEE++ to Grid Bridge”.

The experi
ences were
a little bit disappointing
. The response time
was

much longer

(up to 1 minute

longer) than in
the case of the original Web Service base database service. But in case of some SOAP
message
s the system just hanged or returned with some error messag
es (about incompatible
data types).


Furthermore,

this would be a dead end for the development, since later we cannot extend our
system with any Glob
us related solutions, like grid
-
style certificate
s, resource properties,
factory
-
service
, notification

etc.


6

Outlook


By our developments
, “Grid
-
Enabled SEE++”

may become an efficient tool for supporting
and improving the medical treatment of strabismus.

In the first phase of the Austrian Grid,




w
e e
stablished an i
nitial grid en
abled architecture
based on Globu
s Toolkit 4
for
“Grid
-
Enabled SEE++”

and then we refined it several times
,




w
e parallelize
d

and speeded up the Hass
-
Lancaster Test by a factor of 14
-
17
,




w
e des
ig
ned a grid
-
enabled Patho
lo
gy Fitting algorithm

and




we d
eveloped a Web Service based medical

database for SEE++
.


Based on the results and experiences which we have

gained
,

our ongoing research
(in the
frame of second Phase of the Austrian Grid)
work has

the following directions:




In the future
, we

want

to finish

the WS
-
GRAM extension of “
SEE++ t
o th
e Grid
Bridge” used by C WS Core

and the C WS
-
GRAM
API
s
.

We really think
, it would be
useful

to
make experiments with these APIs and to

report the outcome of t
he
comparative benchmarks
of the usage of
the pre
-
WS and the WS
-
GRAM APIs.




Since we met seve
ral d
ifficulties and limitations in
Globus C WS APIs which
prevented us to finish the development of a fully functioning client for our WSRF
based

database service,

we want
in the future
to search for and investigate other grid
-
enable
d database access tool
s whose

capabilities fit to our requirements

(one option
would be to
use a gLite
-
compatible version of SEE++
, see below
)
.





Another planned way for establishing a grid
-
based distributed medical database is to
use the Grid S
eamless Data Access Middleware (G
-
SDAM) [G
-
SDAM, 2005]

A
USTRIAN
G
RID





14
/
16

developed by the Institute for Applied Knowledge Processing (FAW). The G
-
SDAM
is still under development and the first prototype will come out in September 2006.
However, the developers of G
-
SDAM and SEE
-
GRID have already started to
e
laborate the common requirements and to design interfaces to combine the two
software components.




We are going to implement a variant of pathology fitting that applies a grid
-
based
parallel searc
h technique to find cases in a

distributed medical

database

of

SEE++
that
are similar to measured patient data; using these cases as starting points, the method
executes multiple independent pat
hology fitting processes on the
grid.




Later we plan to extend the SEE
-
GRID medical decision
support system by a surgery
f
itt
ing algorithm that is able to give suggestions

to the doctors for the optimal
treatment of patients.



W
e

have
also joined the EGEE2
project
[EGEE, 2006]

and

designed a
gLite
-
comp
atible

[gLite, 2006]


version of SEE++.

According to this, we intend to fu
rthe
r develop the “Grid
-
Enabled SEE++” software
system on the basis of the higher services of the EGEE2
middleware (compared

with the low
level services of

the Globus Toolkit).

In
[SEE
-
GRID,
2006/3]
, we have already presented the

planned

design of
a

new ar
chitecture of SEE++ based
on gLite 3.0
. Accordingly:




Our next step is to migrate
and integrate
the already existing components of our
software system
into the gLite architecture
.




W
e
intend to modify the starting procedure of SEE++ server processes and we

would
like to
investigate and exploit the interactive
job submission feature of gLite
.




W
e may exchange

the software architecture and
authentication methods applied earlier
for the
“Grid
-
Enabled SEE++”
medical databases to an
AMGA
-
based solution.
AMGA [AM
GA]

is a database access service for grid applications, which is part of the
late
st release of gLite. It is able
to hide the differences of the user interfac
es of the
supported underlying
database systems and provides a unified acc
ess to them with the
grid

style
certificate
-
based authentication

and encrypted network transfer
.




W
e plan to apply R
-
GMA information

system in gLite

in order that

t
he “Grid
-
Enabled
SEE++”

will be able to discover automatically the available databases and the
executer jobs on the g
rid.




A
n
other

important issue
is
manag
ing

Virtual Organizations (VOs),

since we need to
co
ntrol
the access to some

kinds of grid resources.
For instance,
t
he authentication and
authorization of the data sources on the grid are critical

issues for “Grid
-
Ena
bled
SEE++”, because we have to be sure that the

published medical data will be hosted
only by some certain trusted grid nodes.



Finally we

would like

to implement a

variant
of “Grid
-
Enabled SEE++” which uses the Web
Service based interfaces of gLite serv
ices (e.g.: MWSproxy and SOAP
-
based interface of
A
USTRIAN
G
RID





15
/
16

AMGA) and to compare its per
formance with the other “Grid
-
Enabled SEE++”

version based
on the Globus Toolkit 4
.


7

Acknowledgements


The G
-
SDAM framework is developed by the Institute for Applied Knowledge Pro
cessing
(Institut F
ü
r Anwendungsorientierte Wissensverarbeitung ― FAW) as a partner of the SEE
-
GRID project.


References


[AGRID S. Spec., 2005
]
Austrian Grid Software Specification,
http://www.austriangrid.at/austriangrid/internal/deliverables
/docs/WP_I
-
1
/2004/AG
-
DI
-
1
-
2_v0.5
-
draft
.
doc


[AMGA] AMGA User's and Administrator's Manual

http://project
-
arda
-
dev.web.cern.ch/project
-
arda
-
dev/metadata/downlo
ads/amga
-
manual_1_2_3.pdf


[Bosa
,

2006] Karoly Bosa, Wolfgang Schreiner, Michael Buchberger, Thomas Kaltofen.
SEE
-
GRID, A Grid
-
Based Medical Decision Support System for Eye Muscle Surgery
, 1st Austrian
Grid Symposium, December 1
-
2, 2005, Hagenberg, Austri
a. OCG Verlag, 14 pages.


[Buchberger, 2004] Michael Buchberger.
Biomechanical Modelling of the Human Eye
.

Ph.D. thesis, Johannes Kepler University, Linz, Austria, March 2004.

http://www.se
e
-
kid.at/download/Dissertation_MB.pdf



[EGEE, 2006]

E
GEE
-
II homepage, 2006.
http://www.eu
-
egee.org


[Ethereal, 2004] Ethereal Network Protocol Analyzer.
http://www.ethereal.com


[gLite, 2006]

gLite

3.0.0 home page,

2006.
http://www.glite.org


[Globus, 2006] The Globus Toolkit.
http://www.globus.org/toolkit/


[glogin, 2004]
Herbert Rosmanith and Jens Volkert "glogin
-

Interactive Connectivity for the
Grid" in: Z. Juhas
z, P. Kacsuk, D. Kranzlmüller, "Distributed and Parallel Systems
-

Cluster
and Grid Computing", Proc. of DAPSYS 2004, 5th Austrian
-
Hungarian Workshop on
Distributed and Parallel Systems, Kluwer Academic Publishers, Budapest, Hungary, pp. 3
-
11
(Sept. 2004).

http://www.gup.uni
-
linz.ac.at/glogin/


[G
-
SDAM, 2005]
A Report on a Unified Grid
-
aware Access Layer for SEE
-
GRID Data Sets
,

Austrian Grid Deliverable M
-
4aA
-
1c, FAW Institute and RISC Institute, Johannes Kepler
University, Linz, August 2005.
http://www.fa
w.uni
-
linz.ac.at

A
USTRIAN
G
RID





16
/
16


[gSOAP, 2005] gSOAP 2.7.0 User Guide, 2005.
http://www.cs.fsu.edu/~engelen/soap.html


[Kaltofen, 2002] Thomas Kaltofen.
Design and Implementation of a Mathematical Pulley
Model for

Biomechanical Eye Surgery
. Diploma thesis, Upper Austria University of Applied
Sciences, Hagenberg, June 2002.
http://www.see
-
kid.at/download/Pulley_Model_Thesis.pdf


[Mitterdorfer, 2
005] Daniel Mitterdorfer.
Grid
-
Capable Persistance Based on a Metamodel
for Medical Decision Support
. Diploma thesis, Upper Austria University of Applied Sciences,
Hagenberg, July 2005.


[Rosmanith, 2006] Herbert Rosmanith, Peter Praxmarer, Dieter Kranzm
ü
l
ler, Jens Volkert.
Towards Job Accounting in Existing Resource Schedul
ers: Weaknesses and Improvements
,
The 2
nd
International Conference on High Performance Computing and Communications
(HPCC
-
06), Munich, Germany, September 13
-
15, 2006.


[SEE
-
GRID, 2006/3]

Karoly Bosa, Wolfgang Schreiner, Michael Buchberger, Thomas
Kaltofen.
A

R
EPORT ON THE
F
IRST
P
ROTOTYPE OF A
G
RID
-
ENABLED
D
ATA
M
ANAGEMENT
S
YSTEM
FOR
SEE
-
GRID
.

Austrian Grid Deliverable A1c
-
2
-
2006
, Research Institute for Symbolic
Computation (RISC), Johannes Kepler University, Linz, July 200
6


[SEE
-
KID, 2006] SEE
-
KID homepage, 2006.
http://www.see
-
kid.at