Data backup for all - from laptop to supercomputer

Since 1996, the LRZ has operated a central backup and archive system based on the IBM Tivoli Storage Manager software. What started with 2 servers, a small automatic tape library and 4 tape drives has now grown into a huge system consisting of 20 servers, 5 tape robots with over 60,000 slots, 126 tape drives and more than 2,300 hard disks. In addition to the daily backup of thousands of computers from the MWN, archive data from a wide variety of scientific disciplines that need to be stored for the long term are also securely stored by the LRZ. In the following, we would like to give you a small insight into what the LRZ backup and archive system can do. Looking back at the last eighteen years of IT history, one can observe that many companies and with them hardware and software products have come and gone. One of the few software products that has survived this long period - from an IT perspective - is the Tivoli Storage Manager data backup software from IBM used at the LRZ. Of course, the requirements for data backup have changed constantly over the last decades and so TSM has also adapted to these changes in order to meet the requirements. Since the basic strategies of TSM - the so-called Progressive Incremental Backup procedure and the time-controlled retention of data in the archive - are correspondingly flexible, this has always been possible without any problems.

To protect laptops, workstations or servers against data loss due to hardware or software defects or even human error, you can use TSM to create a backup of your files. In this process, new or changed files on your system are detected at regular intervals and transferred to the LRZ where they are stored. All you need for this is an LRZ ID with TSM authorization and a small piece of software, the TSM client, which you have to install and configure on your computer. According to our default settings, the last 3 versions of a file are kept in an LRZ backup and can be retrieved if necessary, for example if you have deleted files by mistake.

When running server applications, especially databases, it may be necessary to actually stop the application during a backup to ensure a consistent backup of the application, since a file must not change while it is being backed up. Of course, this procedure is often not an option. Therefore IBM offers special TSM clients for the following applications, with which such an application can be backed up consistently during operation:

• Databases
• Oracle
• DB2
• MSSQL
• SAP HANA
• Email
• Exchange
• Lotus-Notes

For OpenSource databases like MySQL or PostgreSQL there is the possibility to either create an export of the database, a so called dump, before the backup or to send the dump directly to TSM with the IBM tool adsmpipe without caching the dump locally.

With the triumph of server virtualization, new problems and opportunities for the data protection of these virtual environments have also arisen. Of course it is possible to install a TSM client on every virtual machine (VM) just like on physical ones. However, there is now a special TSM client for the VMware ESX virtualization software that can be used to back up the virtual disks of a VM at the block level from the physical host. With such a backup it is then possible to restore the whole virtual hard disk as well as single files of the virtual hard disk. Furthermore, it is possible to start a VM directly from the backup and then restore it to the production memory of the VMware system during operation via VMotion. This means that the system is already running while it is still being restored in the background.

 In addition to the pure protection of computers against data loss by means of a backup, TSM also provides an archive function, which is used for the long-term secure storage or retrieval of data that no longer changes. By default, each file is stored in the archive system for ten years and automatically deleted after this time. On request, it is also possible to store archive data without a time limit. 

A typical use case for archiving is that data was generated during an experiment that can no longer be stored locally after the evaluation has been completed because the resources are needed for new experiments. But since you can't destroy the data either, because they have to be kept for at least ten years as part of good scientific practice, or because you might have a new idea for evaluating the data at a later time, the data still have to be kept somehow. For this purpose, they can be outsourced to the LRZ archive system.

Since we assume that archive data is deleted from your local system, we make a duplicate copy of all archive data at another location. This ensures that your data can be safely restored, even if storage media or larger areas of the LRZ should suffer damage.

 A certain volume of data in the LRZ backup and archive system has been available free of charge to our statutory users at LMU and TU Munich as part of the basic service since 1996. In 2014, this contingent amounts to 10 PB storage capacity each for LMU and TUM. Over the past 18 years, we have always managed to provide sufficient capacity as a free basic service. We are making every effort to ensure that this will remain the case in the future. 
 
 We offer our services at a preferential price to the other Bavarian universities and scientific or science-related institutions of the Free State of Bavaria. 

If you are interested in our backup and archive services, please feel free to contact us with your questions at any time via the LRZ-Servicedesk. We look forward to receiving your inquiry.