1. Grid simulation (AliEn) Eugen Mudnić Technical university Split -FESB

2. 2 Outline  GRID simulation  Simulation tool – Ptolemy (Berkeley)  Simulation architecture  Simulation software environment  Some results

3. 3 GRID simulation  Why GRID simulation: 1. Validating grid architecture and scalability 2. System resources planning (storage size,network throughput, processing power) 3. Simulating different job placement and file transfer scheduling scenarios 4. GRID services tuning 5. Designing and testing of user strategies (how to efficiently use the system)

4. 4 Modeling and simulation  Top-down or Bottom-up ?  Simulated system is very large and complex (and still evolving)  Appropriately high degree of abstraction  Focusing on the Grid key components Storage elements Network Computing elements  Method : Discrete event simulation (DES)

5. 5  Ptolemy Classic (Berkley) Pros:  Fast (C++), stable  Good DE library (component and messaging system)  Free, source code available Cons:  Complicated (CPP,TCL,PL(Ptolemy language),…)  Resulting simulation needs Ptolemy installation  Ask other Ptolemy users DES tool - Ptolemy

6. 6 C++ object definitions sim. control object construction connecting C++ Ptolemy customization TCL Script (simulation control, object construction, connecting) PL (object definitions) C++ preprocessing compiling obj Ptolemy libraries obj PL (object definitions) PL (component definition) DE engine C++ source

7. 7 AliEn DES model & DE components server CE Network file transfer simulation Storage FTD Message dispatcher Job generation (simplified JDL) FILE catalog

8. 8 Network file transfer calculation component Network as a set of links shared by changeable number of data streams SE1 SE2 SE3 SE4 100 MB/s 30MB/s 50 MB/s 30 MB/s 1 GB/s 100 MB/s LAN1 LAN2 LAN3 links with capacity ( C1,..CL ) N streams, every stream has a predefined transfer route ri  {1,…,L} Every stream has equal priority Stream bandwidth allocation must conform to:

9. 9 Network Storage elements Level 2 tape units RW-access times, throughputs Level 1 (HDD) Cache (CLC) MB/s CE MB/s MSS

10. 10 Software environment AliEnSim simulation application (Ptolemy C++) MySQL database File catalog (directories) Simulation results system state dump Java GUI (system configuration editor, job description, visualization) System configuration LDAP Simulation results & System state dump XML file Results GRID system configuration File catalog operations

11. 11 ¸Java GUI – grid configuration

12. 12 Java GUI – job description

13. 13 Java GUI – grid visualization(control)

14. 14 Results example (jobs running)

15. 15 Some performance testing results  Simulation of Alice PDCIII-Phase I: 104000 jobs execution, output is 3.8mil files (36 per job) 15 sites (total 2100 CPU) central storage at CERN::Castor(MSS) & CERN::scratch  Simulated real time: 10days  Simulation time (P4-2GHz): ~1h

16. 16 Future work ?  Test and enhance component models Identify and use different system measuring  Network – throughput  SE (MSS) : throughput, R/W access times  Optimize,optimize,… Faster real system (more computing elements) slower simulation  Simulate and compare with Alice data processing PDCIII phase I -phase III