Max Planck Institute for Evolutionary Anthropology (EVA)

Story Details

Studying human evolution is more interesting today than ever. Thanks to new computer technology that can create superb 3D images and models of our ancestors from simple bone fragments found in the ground, the story of the development of humankind has really come to life.

The Max Planck Institute for Evolutionary Anthropology (EVA), in Leipzig, Germany, is a leader in the research of human development. Specializing in the analysis of genetics, cultures, languages, and social systems, the institute unites a number of scientific disciplines divided into separate departments. It has published some amazing work — most recently the complete genome of a 38,000-year-old Neanderthal man.

Look more closely at EVA and its teams of scientists capturing the headlines and you’ll see a group of dedicated system administrators in the background carefully maintaining the Institute's network. Just like the investigations into human evolution, the work to ensure that the infrastructure can support the Institute's research never stops.

The demand for data storage has spiraled upward in recent years due to the latest developments in imaging software. The Institute’s 3D-imaging technology, for example, needs quick access to an enormous amount of data when it reads computed tomography (CT) scans and converts them into fascinating, insightful images. Michael Gasch, system administrator for the Department of Human Evolution, says, "The CT scans hold 100 gigabytes of stored data easily. A scientist’s workstation needs 100 MB/s to process the CT data without any noticeable delays.”

In the middle of 2005, the Institute’s Department of Human Evolution chose a Sun StorageTek FLX280 enterprise storage system ahead of competing solutions, initially with 28 TB capacity then upgraded to 81 TB within nine months. After configuring in RAID arrays and deducting storage capacity for reflections, only 35 TB of net capacity remained. Both the administrator and researchers were thrilled with the stable and high-performing storage solution. "The price and performance ratio also matched perfectly," recalls Gasch.

In 2006, the system hit capacity, and the EVA began looking for a more powerful storage system. This time, however, the requirement was for more than simply data capacity and speed. Gasch was looking to move to an online backup process. He also wanted a storage platform that was powerful, problem-free, and complete with SATA drives.

Sun again delivered a winning solution in the form of the mid-range Sun StorageTek 6540 array. These storage arrays are based on LSI technology and have been specially designed with large databases and high-performance computing in mind. They use the new Switched Bunch of Disks (SBOD) technology that controls all drives in an array in parallel for maximum performance.

The new storage arrays were installed by Sun Advantage Partner, Controlware, in record time. Implementing the StorageTek 6540 array including build, configuration, and training sessions took barely a week, and installation was completed in a single day. Formal training sessions were not necessary because of the homogeneous environment — Gasch already knew the process with Sun storage systems from personal experience with the StorageTek FLX280 system.

Gasch opted to use the StorageTek 6540 array with SATA hard drives. Still, because the LSI controller technology allows SATA drives and Fibre Channel to sit inside the same housing, he can switch to the faster Fibre Channel drives in an emergency. The initial tests, even without optimization using a third-party cluster file system, proved successful in producing more than 80MB/s performance. In fact, the figure of 100 MB/s that workstations need to process CT data was easily achieved with the finished configured system.

The Department of Human Evolution has also gained a considerable amount of data protection and availability. The new storage solution has allowed the EVA to save data on two storage arrays. The arrays mirror their files, each host-based on the other device. The two systems are in different computer centers on the campus and are connected via a redundant SAN with two fabrics for business continuity. The other elements of the SAN are also designed to be redundant; together with the LSI failover drivers, three Brocade switches per fabric ensure that there is always a second path to the data.

Depending on availability requirements, the 176 hard drives currently being used in the StorageTek 6540 arrays are configured in RAID 1 or RAID 5 arrays. Dividing them like this does not need to be a final solution. The LSI controllers allow RAID levels to be migrated in the current operation, for example, from RAID 1 to RAID 5.

Since the institute implemented Sun StorageTek 6540 arrays, data storage continues to be problem- free. "Look, in an emergency, we could survive a day without data," says Gasch. "However, after a day, things would become very uncomfortable because not only would the scientific work grind to a halt, but also the entire administration for the department would stop." He adds that thanks to the Sun solution in place, being without data for a day has not been an issue.