Data Service Centre forced to perform well

As an internal service provider of the ÖBB the Data Service Centre fulfils several tasks that are vital to the company. I manages the company's accounting as well as the cost- and combined financial statement in addition to the agenda of 3.400 connected SAP R/3 users.

Extremely high demands are made on the applications of the business administration. There is a considerable amount of data to be processed. For 120.000 already retired employees of the ÖBB  the statements are made by SAP. Consequently, the availability of the service centre is a positive side effect.

Another principal point of tasks urgently points into the direction of SAN / Storage Area Network. Due to the goods traffic and passenger service a 7x24-running is absolutely indispensable. Without a permanent EDP connection no single transport could be possible nowadays. Dealing with orders, which requires highly performing and equally well available systems for its storage, is partly done via the Internet.

To guarantee a stand-by of the system for eventual catastrophes a second data service centre is used.  If environmental catastrophes, terrorist attacks or software, respectively hardware upgrades lead to a standstill in the operative Data Service Centre, a switch onto a second computer site can be undertaken. In this way a complete breakdown is to be avoided.

ÖBB expands SAN

The ÖBB started the introduction of  SAN as early as in 1999. The first tests were launched in 2000. Very soon the full operation of first servers followed. The extremely rapid growth of SAN led to an early reconsideration of the overall concept of storage capacity. The result of these considerations was a convincing solution. Instead of Hubs and Point-to-Point connections Switches should meet their difficult demands.

By mid 2000 the first high-performance directors were in use in the Fibre-Channel. By the end of 2000 the ÖBB had the first larger Data Service Centre within Europe to operate such a very remarkable 64-Board-Switch. Since then, the ÖBB have noted such a dynamic growth, that their Data Service Centre can include up to 50 additional servers.

All in the SAN - a status for the future

Now nearly all of the altogether 160 servers of the Data Service Centre are connected to the SAN.  All discs run in the Storage Area Network, so that, with the exception of the Boot-Area no external discs are necessary. Mag. Wolfgang Danzinger, manager of the project, explains: "Our concept is based on the idea of providing a completely switched environment. Our next step will be to make the SAN redundant enough to prevent any Singe Point of Failure from interrupting the system."

Hard discs and tape processing

To continue, the ÖBB want to join their decentralized servers. File- and printservers are consolidated in the Data Service Centre. Solutions for storage will be located in the Data Service Centre as well. As to fileservers and storage systems there will be less confidence in jukeboxes than in the so called NAS (Network Attached Storage).

COMPAREX, partner of the German firm GRAU is placing more trust in their INFINISTORE Virtual Disk (IVD).

This IVD can run worm-cartridges. These are without any doubt legally accepted media for storage, comparable to conventional CDs or Cartridges for jukeboxes.

The advantage of such Virtual Discs is their nearly unlimited scalability. Besides, IVD-systems can be assigned to a number of servers, whereas a jukebox could only be assigned to one server. This improves the performance and scalability. After all, IVDs show the same high functionality as central Plattenhubsysteme, and so they offer Remote Copy, Spiegelung und Snapshots, failure security with 2 sites and a lot more. 

Fileserver consolidation in Insourcing

The fileserver consolidation considered as the first step for the ÖBB Data Service Centre plans the insourcing of logically 60 servers into the Data Service Centre.

In order to handle the expensive technology of discs in a relatively economic way in spite of the enormous amount of data, parts of the data shall be kept on the disc , whereas other parts of the data shall be kept on cartridges in the IVD-system.

At the same time all of it is depicted in a virtual system, so that it is not of any importance to the user where the data are stored. Obviously, the use of the cheap medium of cartridge  will have considerable economic effects which multiply by the factor 10 for file-servers", explains Mag. Wolfgang Danzinger of the ÖBB.

Older data, which are only rarely retrieved, can be stored in the cheapest way due to this concept. Of course they are as well available to the user, as if they were stored on discs. Presently, the IVD is used for storage. Mr. Danzinger wants to start this filserver consolidation as soon as possible. Then HSM - functionality (Hierarchical Storage Management) will be available also there.

7.5 times the amount of storage manageable

IDC studies prove, that central solutions like SAN or other central storage solutions make it possible for the same number of employees to mange up to 7,5 times the amount of data. These IDC studies have been verified in the practical experience of the ÖBB. " We still have only three employees to manage already 13TB of data.

Previously, we only had to manage 300GB", Mr. Danzinger confirms. " So our positive financial expectations were also met concerning the criteria of economy. It was another important aspect for us, that staff-related costs should only amount to about 15% of the total peripheric costs. This as well has turned out achievable and right. "

Partnership with a future - ÖBB & COMPAREX

The further progression of the project will bring a complete realisation of the Failover-concept. The ÖBB rely on COMPAREX's accompanying help and support. An extensive partnership of this kind is necessary and meaningful for the ÖBB, as  the surroundings are new both for the market and for the ÖBB themselves.  

"At the moment our SAN seems like a Single Point of Failure" Mag. Wolfgang Danzinger , who is responsible for the ÖBB project, states. " If one component - starting with a cable up to a switch - happens to fail within the SAN, the server concerned is at a standstill. But as all servers are organized in a redundant way, the SAN also has to meet this requirement." Therefore a lot of the redesign in storage network will become reality. At the same time the distribution of load and the measurement of performance will be integrated. So it is left to be a great challenge for the storage experts of COMPAREX.