Dealing with Long-Lived Data in High Performance Storage Systems

Long-term reliability and manageability of high-end storage systems for scientific computing will become an increasingly important issue as disk-based storage systems grow to handle petabytes and, eventually, exabytes of data. Storage systems cannot simply be upgraded by replacing them; instead, they must grow over time. Managing and replacing the thousands of individual devices that make up such a system, marshaling the trillions of files stored in the system, and ensuring the security of the data they contain will pose significant challenges that must be solved to ensure that data stored in the system today can be found and used in 25 years.