ALICE    A Large Ion Collider Experiment
Hadron Physics Department - IFIN-HH
Contribution to ALICE Experiment @ LHC
Data center
The "project" started in 2005, when appropriate funding become available
Distributed computing - 2005
NFS regularly used between user computers
MOSIX cluster
AliEn site
Distributed computing - past
HPD members regularly used clusters at international sites

Our group is involved in Grid Computing activities.
Starting from November 2002 the Computing Cluster of our Centre of Excellence is included in the ALICE Grid Structure, using AliEn Grid Environment, like a Computing Element (CE), becoming the first international Grid application in Romania. The very first CE containde about 10 CPUs (AMD Athlon MP and Intel Pentium processors) with a total computing power of 5588 SI2k, and a total disk storage capacity of about 200GB. The communication between computing nodes was at 100Mb/s and the link to the Internet at 100Mb/s. At that time this configuration was used to run ALICE test jobs and Nuclear Structure large scale model calculations. Soon after this starting period the computing power was upgraded with about 40 CPUs, using dual machines based on Intel Xeon and AMD Opteron CPUs. The Storage Element (SE) for the ALICE Grid was increased by about 2TB and the link to the Internet provider at 1Gb/s. The final target being to have a Tier2 Centre by the time when the first experimental information will be delivered by the ALICE experiment at CERN.
2006 takeoff
~ 200K CPU hours
(~300 kSI2k hours)
19 Kjobs DONE!  - 6,8%
Last 4 months 11 TB saved @CERN
Average speed ~1MB/sec
Maximum speed ~48MB/sec
99,5% availability
The first production-grade cluster, 1 frontend machine (dual Xeon 3GHz, 2GB RAM, 2.4 TB raw HDD), 6 nodes (dual Xeon 3GHz, 4GB RAM) another 4 at the end of the year, all server-class, 32bit
1 Gb/s network
Fortunately, we had a proper space within the Detector Laboratory, cooled by the Lab s unit
2007 developments
- 15 new machines (dual dual-core Opteron, 2GHz, 8GB RAM)
- 48 TB raw SAS storage (new technology)
- 10 Gbit/s connection to DIC
- 40 new machines deployed during the year (dual Xeon 3.2 Ghz, 2 MB L2, 4GB RAM), 64 bit
- EGEE site
- Policy1: "Regarding GRID, there is nothing more important than having a running and used site"
- Policy2: try to find as fast as possible solutions for the problems showing up during production
- Policy3: exploit to the maximum extent the "dedicated" character of the site in order to achieve high stability and availability
~ 1000 kHours (delivered)
~ 130 Kjobs DONE
~ 4-6% of ALICE cumulated for last 2years
~ 3360000 files saved to NIHAM SE
Average Site Services Availability
since Iuly 1, 2006 :>95%
~33500 jobs DONE
~360 kHours CPUTime
~7% of ALICE
Starting with September 2006, NIHAM
Storage Element was used by ALICE production jobs to store log files
Among the firsts AliEn2 sites
In production since september 2005
838 jobs done
The most important development: NIHAM Datacenter, industrial grade cooling unit, industrial-grade UPS-es, Cat. 6 cabling, connection to Institute's Diesel generator.
600 kW
- control and monitoring
         via ethernet
Modern Diesel generator
Computing power: 36 servers
                   Storage: 200 TB - HDD

the increase of the software efficiency and performance of the
    new processors shown evidenced an insufficient storage capacity
    of NIHAM Data Centre
A unbalanced computing power vs. storage capacity has as
                 -  overload of the existing storage servers
decrease of their performance
network overload
lower efficiency in using the existing computing power
A back-up computing power in case of interventions
An upgrade of NIHAM network for supporting the present needs and developments
Extension of the storage capacity ~ 500 TB
Computing servers - 10 servers
Extension of the local network infrastructure - 4 switches & accessories 
2010 - 2011
~ 2000 CPU cores

~ 1 PB storage
New cooling unit

- temperature
- humidity
- fresh air