R ti f S untime for S Runtime for S Si l ti W kfl SimulationWorkflows

Transcription

R ti f S untime for S Runtime for S Si l ti W kfl SimulationWorkflows
IIntegrated
t g t d Data
D t Management,
M
g
t, W
Workflow
kfl
and
d Visualization
Vi
li ti
t Enable
to
E bl an Inte
I tegrative
g ti Systems
Sy t
S
Science
i
PN 8
R ti
Runtime
for
f S
Si l ti Workflows
Simulation
W kfl
Prof.
P
f D
Dr. rer. nat.
t F
Frank
kL
Leymann
y
Dipl Math.
Dipl.-M
M
Math
Michael Reiter
Geophysical
en ironment of
environment
ground,
d air,
i and
d
water
Globally measured data
and observations over a
particular
ti l time
ti
period
i d ((e.g. Collect and
models of
process Analysis and Mathematical-physical m
6 h)
numerical weather fo
orecast
d t
data
assimilation
PN 8 develops an integrative simulation environment supporting
different scientific domains.
domains The environment supports modelling of
process logic of scientific experiments,
e periments their execution
e ec tion and thus
th s
simulation.
i l ti
DMO
(
(direct
model
output)
p )
Statistical
St
ti ti l
i t
interpretation
t ti
Legend
Acti it
Activity
Weather
forecast for
users
Subjective
interpretation by
meteorologist
t
l i t
Automated and distributed execution
Simplified integration of IT infrastructure (databases, servers,
Statistical pos
p stprocessing
p
g
(e.g. using
KALMAN filterrs)
Control flow
text
The use of workflow technology
gy allows:
Activity
d
description
i ti
sensor nets, …)
Reuse of existing algorithms
algorithms, scientific experiments and
simulations
Robustness of simulation execution
6,4 l/m²
l/m
Fi
Figure
1
1: S
Scientific
i tifi workflow
kfl
ffor weather
th forecast
f
t
Projecct Subject
Runtime infrastructure for Simulation Workflows
Standard based execution infrastructure for simulation workflows
C
Conventional
ti
l workflow
kfl
t h l
technology
adapted
d t d for
f scientists
i ti t
Existing features: reliable,
reliable available,
available extensible,
extensible scalable,
scalable secure,
secure tran
nsactional etc.
nsactional,
etc
Interoperable with heterogeneous and distributed simulation software
R
Reusing
i
existing
i ti simulation
i l ti software
ft
Reproducibility of simulation: other scientists can reconstruct the simu
ulation and the results
Optimized usage of resources like software, servers (e.g. supercomputers), databases, etc.
Run
Ru
ntiim
me
e
Si l ti Workflow
Simulation
W kfl w Management
M
t System
S t
Figure 2: Architecture of the Simulation Workflow Management System (WfMS
S)
Figure 3: Web Service interface with Dune and ChemShell plugin
Si l tion Examples
Simulatio
E
l
Simulation Services
Dune‐
based
Si l ti
Simulation
Workflow
Figure 4: Heat conduction experiment in reality
Figure 5: Result after 90 minutes for a 3D
transient heat cond
duction simulation
that can be used with the SimTech WfMS:
G
Generic
i simulation
i l ti preparation
ti
Numerical solving: Dune
diffusion
h t conduction
heat
d ti
Chemical reactions: ChemShell
glutamate mutase
Porous media: Pandas
coupled problems
I i
Institute
Institute of Architecture of
off Architecture
A hi
offf Application Systems
A li i Systems
Application
S
Katharina Görlach Dimka Karastoyan
Katharina Görlach, Dimka Karastoyan
nova Frank Leymann Michael Reiter Mirko Sonntag
nova, Frank Leymann, Michael Reiter, Mirko Sonntag