Development of the Computing Centre Rossendorf

1961: Foundation of the ZfK Computing Centre. Acquisition of the ZRA1, because of several problems it started running to the full extent in May 1963. Afterwards the employees dominated the numeric evaluation of complete gamma scintillation spectrums with a program on the ZRA.

1962 up to 1963: Start of the development of the "Rossendorfer subprogram technique" by employees of the Computing Centre. A pool with more than 400 subprograms was developed for the subprogram technique.

The thermionic analog processor endim 2000 from Glashütte was set up in the Computing Centre and supplemented by a second at the end of the year.

 

1964: Start of the use of the computer NE 503 from the company National Elliott, which was positioned in Dresden because of physicians from the ZfK with the support of the Computing Centre.

1966 up to 1969: Development of a new hardware (ZRA2) and the appropriate software (including system software) as well as an advancement of the “Rossendorfer subprogram technique" with the development  of an ALGOL Compiler. The delivery of the ZRA2 for a general use was in January 1969.

 

1975: Software operations for data processing with the TU of Dresden and the Computing Centre of Berlin.

An optimized system of four virtual engines was realised on the ZRA2.

The data center is situated in building 13.

 

1978: Start of the construction of a new building for the ESER processor.

1981: Delivery of the new building to the employees of the Computing Centre.

1983: Implementing of the EC 1055 at the Computing Centre Rossendorf.

,Gebäude 13c - Rechenzentrum 2006

1983-1991: Enlargement of the EC 1055 at the Computing Centre Rossendorf. The picture derives from the late eighties when new user terminals were introduced. In 1991 the reduction of the EC 1055 processor took place to make way for new and modern techniques.

1991: Accession to the research net (WIN, Internet) of Germany.

Rechenzentrum Ende der Achtziger

The installation and implementing of the Convex 3220 took place in December 1991.

1992: Foundation of the Forschungszentrum Rossendorf (FZR).

1994:  Enlargement of the Computing Centre with a CONVEX 3820 (2 vector processors, 1 GB Memory).  The older CONVEX 3220 accepts the role as a file server. In contrast to the older one  stands the new CONVEX 3820 by resource intensive jobs .

1995 up to 1996: The World Wide Web was opened up for the FZR.

Because of the installation of a tape archive new dimensions of safety and administration were opened. The research net (WIN) was upgraded from 64KBit/s to 128 KBit/s.

1997 up to 1998: The CONVEX processor was substituted by a modern simultaneous computer HP V2250 (8 processors at 240 MHz).The B-WIN research net was enlarged from 128 KBit/s to 2 MBit/s.

1999 up to 2000: A second simultaneous computer HP V2250 with 16 processors was installed and the already available simultaneous computer was enlarged from 8 to 16 processors. The central archive and backup services were expanded and substituted by a new archive processor called HP K570. A new standard PC was defined in cooperation with the users. The employees participated in the trainee of students with courses and internships.

2001: The WIN accession was converted from B-WIN to G-WIN with 2 MBit/s.

2002: The Linux Cluster with firstly 16 nodes (32 CPU) was started up in February.The implementing of the new centric file server of the FZR (STK D280, 4 TByte capacity, later enlargement to 12 TByte) was in December.

2003: The FZR is added to the G-Win of the DFN. The first Gigabit routes at the FZR backbone are implemented in July.

2004: Extensive enlargements of the archive and backup capacity. Expansion of the IBM LTO tape library at LTO2 Laufwerke (400 GB capacity at tape) and an additional rack with 440 slots. For this reason a capacity of more than 100 Terabyte is available.

Also the Linux Cluster for the applications of the High Performance Computing will be enlarged. In the middle of 2005 it consists of 44 compute nodes with 88 Intel Xeon CPU (2,0-3,6 GHz, 100 GByte Memory).

IBM Bandbibliothek 2007
2005-2006: The technical Infrastructure of our data centre is renewed. The power supply is done by redundant power distrubution units, which are backed-up via a 250 kVA UPS. An emergency diesel generator provides uninterrupted power supply.  In 2006 the first water-cooled computer racks are installed.

The new Linux-Cluster based on dual-Opteron CPU tripples the installed computing power in the data centre.

Step by step more and more server are consolidated on a few powerful hardware nodes and operated as virtual server on the VMware ESX platform.

2007: The data centre gets a little sister: A backup data centre  is established in building 13 to save the data and provide a minimum a set of services for a catastrophic scenario.

HPC-Cluster Hydra am FZD

2008-2010: The massive data growth is continuing, 2009 a new storage BlueArc Titan 3110 for the large-scale data is installed and expanded up to more than 400 Terabyte till 2011! In the communication network the first switches with 10 Gigabit bandwidth are being used.

The virtualization limits the hardware increase for the services. By the of 2010 about 100 virtual servers  run on the new hardware platform consisting of two HP Blade7000 shelfs. 

In the area of High Performance Computings the growth continues, in addition to the cluster Hydra the new cluster Hypnos was added providing  832 AMD CPU-cores and 3,3 Terabyte main memory.

BlueArc Titan 3100

2011: After 30 years of continuous service the "old" data centre is not sufficient with respect to rack area and cooling capacity. A fundamental expansion of the building and installation of new cooling technology is planned and realized. The reconstruction was done in two segments during full operation.

Here you can see a little impression, it looks like being in a waterworks...

Boden beim Umbau des Rechenzentrums
2012: The expansion project is finished, now we have the space for more than 50 racks and a cooling capacity of 400 kW! The cooling is based on an intelligent mixture of free cooling, absorption and compression devices.

We also have expanded our hypnos-cluster with both CPU- and GPU-nodes. Now we can offer more than 7.000 CPU-cores and more than 100.000 GPU cores for the computations.  

At the completion of our advanced data center and our significantly extended supercomputer Hypnos, a scientific symposium was held on 12th of December.

Agenda of the Event

Impressions of the event in a picture Gallery

Rechenzentrum des HZDR

2014: Big Data is the new challange! To handle it a new large IBM storage with 2 Petabyte capacity and a throughput of 15 GByte/s was installed.

The storag is based  on the parallel filesystem GPFS, thousends of Clients can write at the same time. The cluster nodes, connected via Infiniband, have a great benefit.

IBM GSS Storage