Grid Activities in Ukraine

1
Grid Activities inUkraine

• Nataliya Kussul
• Space Research Institute
• NASU-NSAU, Ukraine

WGISS 21, Budapest 2006
2
Main issues

• Infrastructure
• Hardware resources
• Technologies
• Applications
• Numerical Weather Prediction
• Meteosat Data Processing Services
• Cloud Mask
• Cloud Motion
• Fractal Feature Extraction
• Current State of Grid
• Projects involved
• Prospects

3
Current Infrastructure
4
Hardware resources

• Cluster SKIT-1
• 24x2 Intel Xeon (2.6 GHz)
• Infiniband interconnect
• Cluster SKIT-2
• 32x2 Intel Itanium II (1.4 GHz)
• SCI interconnect
• Development cluster
• 2x2 AMD Opteron (2.2 GHz)
• Gigabit Ethernet interconnect

5
Hardware resources

• LustreFS NAS 14TB storage
• iSCSI SAN 2TB storage
• Gigabit Ethernet inter-cluster connection
• A number of UNIX-based workstations
• Condor pool of MS Windows workstations

6
SKIT-1
SKIT-2
Development cluster
7
Cluster load reports

• WRF model load profile

8
Technologies

• Globus Toolkit v4 Grid middleware
• Karajan workflow description language
• UNM MapServer OGC-compliant software for
  geospatial data presentation
• Torque local cluster scheduler
• OpenMPI Scaly MPI standard implementations
• Ganglia monitoring tool
• Ruby scripting language

9
Applications

• Meteosat cloud mask extraction
• Cloud motion estimation
• Fractal features extraction
• Numerical Weather Prediction (WRF model)

10
WRF run subsystem (1)

• Configured for territory of Ukraine
• Input boundary condition from GFS
• 3 day forecasts every 6 hours
• 10 km horizontal grid, 200x200 gridpoints
• 31 vertical levels
• One run takes 3.5 hours on 2x2 Opteron system

11
WRF run subsystem (2)
12
WRF run subsystem (2)
13
WRF visualization

• UMN MapServer based
• Available at http//cosmogis.org.ua/?optioncom_wr
  f

14
Meteosat visualization

• UMN MapServer based
• Available at http//cosmogis.org.ua/?optioncom_ma
  pserver

15
Cloud motion estimation

• useful as the source data for Numerical Weather
  Prediction models
• temporary interpolation/extrapolation of image
  sequences for now casting
• based on adapted optical flow techniques

16
Fractal features extraction

• Based on modeling of Meteosat images by
  Generalized Brownian Motion processes
• Useful for detecting turbulent areas

17
Current state

• At the moment Grid middleware is being deployed
  on SKITs frontend node.
• SKITs are using proprietary local scheduler
• An adapter between Globus GRAM and scheduler
  should be developed

18
Projects involved (1)

• Cosmogis project supported by Ukrainan National
  Space Agency
• STCU project Grid technologies for environmental
  monitoring using satellite data (started on
  December, 2005)
• Innovative project of National Academy of Science
  of Ukraine Development of pilot project of
  Ukrainian GEOSS/GMES segment (2006)

19
Projects involved (2)

• INTAS-CNES-NSAU call
• Project Data Fusion Grid Infrastructure is
  under preparation -
• Practical way to initiate cooperative activity
  of agencies on Grid creation
• We are looking for partners!

20
Plans Technology

• To develop Grid-enabled archive for geospatial
  data based on OGSA-DAI
• To establish data assimilation of MSG and ATOVS
  data into modeling process
• To use Grid infrastructure to perform a model
  automatic adaptation experiment
• To work out patterns for inter-domain security
  policy

21
Plans organizational

• To connect to European academic network GEANT
• To connect to Ukrainian academic network URAN
• To become a regional center of GEOSS/GMES
• To participate in the development of WAG
• To include Ukrainian databases into ECHO system

22

• Thank you!